{"_id":"kagura-memory","_rev":"23-e93b18c4c50c9a2a949c39e068211c13","name":"kagura-memory","dist-tags":{"latest":"0.19.1"},"versions":{"0.1.0":{"name":"kagura-memory","version":"0.1.0","keywords":["ai","memory","mcp","agent","llm","kagura","rag","embeddings","vector-search"],"author":{"name":"Kagura AI","email":"dev@kagura-ai.com"},"license":"MIT","_id":"kagura-memory@0.1.0","maintainers":[{"name":"kagura-ai","email":"fumikazu.kiyota@gmail.com"}],"homepage":"https://github.com/kagura-ai/kagura-memory-typescript-sdk","bugs":{"url":"https://github.com/kagura-ai/kagura-memory-typescript-sdk/issues"},"dist":{"shasum":"5b1f217629549d686d74661afa825918b6d5e505","tarball":"https://registry.npmjs.org/kagura-memory/-/kagura-memory-0.1.0.tgz","fileCount":9,"integrity":"sha512-EHNWdq0cMOX8o8vyV/3K57tHrqeoi4ysCFeQGlKv1NWsnQ9Vcg6snFCw41oaJN/OXcTkIg3MCYD416oaS3QlaQ==","signatures":[{"sig":"MEYCIQDO3YIsqeGM8UpaqazRXpSjjJmGeg3wLSq1b1V/YWulKAIhAM+Y90JgPGNyNigrui3MRxSPw5+Gw4C4g7GkwQ5mGilt","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"}],"unpackedSize":867329},"main":"./dist/index.cjs","type":"module","types":"./dist/index.d.ts","module":"./dist/index.js","engines":{"node":">=18"},"exports":{".":{"import":{"types":"./dist/index.d.ts","default":"./dist/index.js"},"require":{"types":"./dist/index.d.cts","default":"./dist/index.cjs"}}},"gitHead":"6df2ca8b785d182c46d409b3337b46965f34dd12","scripts":{"test":"vitest run","build":"tsup","typecheck":"tsc --noEmit","test:watch":"vitest","prepublishOnly":"npm run typecheck && npm run test && npm run build"},"_npmUser":{"name":"kagura-ai","email":"fumikazu.kiyota@gmail.com"},"repository":{"url":"git+https://github.com/kagura-ai/kagura-memory-typescript-sdk.git","type":"git"},"_npmVersion":"10.9.3","description":"TypeScript SDK for Kagura Memory Cloud — memory management, graph exploration, and R2 file storage for AI agents","directories":{},"_nodeVersion":"22.20.0","_hasShrinkwrap":false,"devDependencies":{"tsup":"^8.3.5","vitest":"^2.1.8","typescript":"^5.7.2","@types/node":"^22.10.0"},"_npmOperationalInternal":{"tmp":"tmp/kagura-memory_0.1.0_1783325077562_0.6846446032512825","host":"s3://npm-registry-packages-npm-production"}},"0.2.0":{"name":"kagura-memory","version":"0.2.0","keywords":["ai","memory","mcp","agent","llm","kagura","rag","embeddings","vector-search"],"author":{"name":"Kagura AI","email":"dev@kagura-ai.com"},"license":"MIT","_id":"kagura-memory@0.2.0","maintainers":[{"name":"kagura-ai","email":"fumikazu.kiyota@gmail.com"}],"homepage":"https://github.com/kagura-ai/kagura-memory-typescript-sdk","bugs":{"url":"https://github.com/kagura-ai/kagura-memory-typescript-sdk/issues"},"dist":{"shasum":"380f634181e1a2592201c090b9978dffa8141388","tarball":"https://registry.npmjs.org/kagura-memory/-/kagura-memory-0.2.0.tgz","fileCount":9,"integrity":"sha512-gem0k3sz+Pp6E7XVFo0Rx38tSiRR+tHczqBsFIl0ygUHiFXK0CrH8u6q6XKmJv5wQMQ9skCf7jquAUJuAC5BKg==","signatures":[{"sig":"MEQCIF9Z+rF92V1LC0Y2FSZN7ibGTA4xzjRGLoHXe7j6PEFYAiA3Y0Z33BCFugjO+qrbk9c690GiAPRqgsgFAKTdLvV/wA==","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"}],"unpackedSize":976866},"main":"./dist/index.cjs","type":"module","types":"./dist/index.d.ts","module":"./dist/index.js","engines":{"node":">=18"},"exports":{".":{"import":{"types":"./dist/index.d.ts","default":"./dist/index.js"},"require":{"types":"./dist/index.d.cts","default":"./dist/index.cjs"}}},"gitHead":"92f7ff197551da8a82b14e9fe99b363bc7db4656","scripts":{"test":"vitest run","build":"tsup","typecheck":"tsc --noEmit","test:watch":"vitest","prepublishOnly":"npm run typecheck && npm run test && npm run build"},"_npmUser":{"name":"kagura-ai","email":"fumikazu.kiyota@gmail.com"},"repository":{"url":"git+https://github.com/kagura-ai/kagura-memory-typescript-sdk.git","type":"git"},"_npmVersion":"10.9.3","description":"TypeScript SDK for Kagura Memory Cloud — memory management, graph exploration, and R2 file storage for AI agents","directories":{},"_nodeVersion":"22.20.0","_hasShrinkwrap":false,"devDependencies":{"tsup":"^8.3.5","vitest":"^2.1.8","typescript":"^5.7.2","@types/node":"^22.10.0"},"_npmOperationalInternal":{"tmp":"tmp/kagura-memory_0.2.0_1784180340991_0.8329329000976224","host":"s3://npm-registry-packages-npm-production"}},"0.3.0":{"name":"kagura-memory","version":"0.3.0","keywords":["ai","memory","mcp","agent","llm","kagura","rag","embeddings","vector-search"],"author":{"name":"Kagura AI","email":"dev@kagura-ai.com"},"license":"MIT","_id":"kagura-memory@0.3.0","maintainers":[{"name":"kagura-ai","email":"fumikazu.kiyota@gmail.com"}],"homepage":"https://github.com/kagura-ai/kagura-memory-typescript-sdk","bugs":{"url":"https://github.com/kagura-ai/kagura-memory-typescript-sdk/issues"},"dist":{"shasum":"f2b53d81ca0d949c0c7061cefaeba136a58dfee6","tarball":"https://registry.npmjs.org/kagura-memory/-/kagura-memory-0.3.0.tgz","fileCount":9,"integrity":"sha512-KcinrYH2pY4vrHCXTulrouytmFACeC1UTzzEp/Tz5/8o4HC7Zp+QppgI1xeZa2oYWjN75ij7ev/zD9o/cc5mCA==","signatures":[{"sig":"MEUCIQCSNl41s/e6QXduAUWgPo/UENZF8ejxHJzuX4HGWwIY4wIgCV3uSIeBouQJibJsyyQnTOtfaxmlpxA7oaJyPb2hC1I=","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"}],"unpackedSize":972369},"main":"./dist/index.cjs","type":"module","types":"./dist/index.d.ts","module":"./dist/index.js","engines":{"node":">=18"},"exports":{".":{"import":{"types":"./dist/index.d.ts","default":"./dist/index.js"},"require":{"types":"./dist/index.d.cts","default":"./dist/index.cjs"}}},"gitHead":"5ce41e4c5e8b61880e20d036815564e0f77be2d5","scripts":{"test":"vitest run","build":"tsup","typecheck":"tsc --noEmit","test:watch":"vitest","prepublishOnly":"npm run typecheck && npm run test && npm run build"},"_npmUser":{"name":"kagura-ai","email":"fumikazu.kiyota@gmail.com"},"repository":{"url":"git+https://github.com/kagura-ai/kagura-memory-typescript-sdk.git","type":"git"},"_npmVersion":"10.9.3","description":"TypeScript SDK for Kagura Memory Cloud — memory management, graph exploration, and R2 file storage for AI agents","directories":{},"_nodeVersion":"22.20.0","_hasShrinkwrap":false,"devDependencies":{"tsup":"^8.3.5","vitest":"^2.1.8","typescript":"^5.7.2","@types/node":"^22.10.0"},"_npmOperationalInternal":{"tmp":"tmp/kagura-memory_0.3.0_1784183262488_0.13340602804990986","host":"s3://npm-registry-packages-npm-production"}},"0.5.0":{"name":"kagura-memory","version":"0.5.0","keywords":["ai","memory","mcp","agent","llm","kagura","rag","embeddings","vector-search"],"author":{"name":"Kagura AI","email":"dev@kagura-ai.com"},"license":"MIT","_id":"kagura-memory@0.5.0","maintainers":[{"name":"kagura-ai","email":"fumikazu.kiyota@gmail.com"}],"homepage":"https://github.com/kagura-ai/kagura-memory-typescript-sdk","bugs":{"url":"https://github.com/kagura-ai/kagura-memory-typescript-sdk/issues"},"bin":{"kagura-memory":"dist/cli.js"},"dist":{"shasum":"c596f260204f42da4dc9e1a5b4ddf9ef10d05717","tarball":"https://registry.npmjs.org/kagura-memory/-/kagura-memory-0.5.0.tgz","fileCount":11,"integrity":"sha512-WCMIDIUOyon3nAJKYqrn2tZUNhm2TWLGZvm2U47nWlW5AOz5sfsbbaRnzNLoQ2qZvl0WeDzrdgjPmWS7wrwlag==","signatures":[{"sig":"MEQCIFzhHZr2j5WJIdX1Xz6ajQ74QxUkNijB3BgODO8A+JsDAiAVtUCuj2aAXrZ0P5Jn8vulPBjbyDGbXCytXznjN0xo1A==","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"}],"unpackedSize":1311086},"main":"./dist/index.cjs","type":"module","types":"./dist/index.d.ts","module":"./dist/index.js","engines":{"node":">=18"},"exports":{".":{"import":{"types":"./dist/index.d.ts","default":"./dist/index.js"},"require":{"types":"./dist/index.d.cts","default":"./dist/index.cjs"}}},"gitHead":"15e1edcf25a155a8d3a09d8db5a1608c599eb5d6","scripts":{"test":"vitest run","build":"tsup","typecheck":"tsc --noEmit","test:watch":"vitest","prepublishOnly":"npm run typecheck && npm run test && npm run build"},"_npmUser":{"name":"kagura-ai","email":"fumikazu.kiyota@gmail.com"},"repository":{"url":"git+https://github.com/kagura-ai/kagura-memory-typescript-sdk.git","type":"git"},"_npmVersion":"10.9.0","description":"TypeScript SDK for Kagura Memory Cloud — memory management, graph exploration, and R2 file storage for AI agents","directories":{},"_nodeVersion":"22.12.0","_hasShrinkwrap":false,"devDependencies":{"tsup":"^8.3.5","vitest":"^2.1.8","typescript":"^5.7.2","@types/node":"^22.10.0"},"_npmOperationalInternal":{"tmp":"tmp/kagura-memory_0.5.0_1785241184854_0.9866651272781335","host":"s3://npm-registry-packages-npm-production"}},"0.6.0":{"name":"kagura-memory","version":"0.6.0","keywords":["ai","memory","mcp","agent","llm","kagura","rag","embeddings","vector-search"],"author":{"name":"Kagura AI","email":"dev@kagura-ai.com"},"license":"MIT","_id":"kagura-memory@0.6.0","maintainers":[{"name":"kagura-ai","email":"fumikazu.kiyota@gmail.com"}],"homepage":"https://github.com/kagura-ai/kagura-memory-typescript-sdk","bugs":{"url":"https://github.com/kagura-ai/kagura-memory-typescript-sdk/issues"},"bin":{"kagura-memory":"dist/cli.js"},"dist":{"shasum":"31bf8affcc649d42682db11c93b062ce17712a73","tarball":"https://registry.npmjs.org/kagura-memory/-/kagura-memory-0.6.0.tgz","fileCount":11,"integrity":"sha512-1ACqTADuoCHYHkm7fdy10eFA597FkMl3YU0uGKa+B7qTDb9BmUCW2LzaGPv5dDPUMQG1nTZ8Wd6+61T+Er7ipA==","signatures":[{"sig":"MEYCIQD7XpkJLnjY44NyhPvgKgbfVEcTJl17NsRF2CAmUHsgrQIhAKOMfDug3lsW8XfXn3s41MnGCY/UFytwrqDf5fpe+A95","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"}],"unpackedSize":1313304},"main":"./dist/index.cjs","type":"module","types":"./dist/index.d.ts","module":"./dist/index.js","engines":{"node":">=18"},"exports":{".":{"import":{"types":"./dist/index.d.ts","default":"./dist/index.js"},"require":{"types":"./dist/index.d.cts","default":"./dist/index.cjs"}}},"gitHead":"48da20857e2010f19ac28d849756453d41840570","scripts":{"test":"vitest run","build":"tsup","typecheck":"tsc --noEmit","test:watch":"vitest","prepublishOnly":"npm run typecheck && npm run test && npm run build"},"_npmUser":{"name":"kagura-ai","email":"fumikazu.kiyota@gmail.com"},"repository":{"url":"git+https://github.com/kagura-ai/kagura-memory-typescript-sdk.git","type":"git"},"_npmVersion":"10.9.0","description":"TypeScript SDK for Kagura Memory Cloud — memory management, graph exploration, and R2 file storage for AI agents","directories":{},"_nodeVersion":"22.12.0","_hasShrinkwrap":false,"devDependencies":{"tsup":"^8.3.5","vitest":"^2.1.8","typescript":"^5.7.2","@types/node":"^22.10.0"},"_npmOperationalInternal":{"tmp":"tmp/kagura-memory_0.6.0_1785250037402_0.8818145865078777","host":"s3://npm-registry-packages-npm-production"}},"0.7.0":{"name":"kagura-memory","version":"0.7.0","keywords":["ai","memory","mcp","agent","llm","kagura","rag","embeddings","vector-search"],"author":{"name":"Kagura AI","email":"dev@kagura-ai.com"},"license":"MIT","_id":"kagura-memory@0.7.0","maintainers":[{"name":"kagura-ai","email":"fumikazu.kiyota@gmail.com"}],"homepage":"https://github.com/kagura-ai/kagura-memory-typescript-sdk","bugs":{"url":"https://github.com/kagura-ai/kagura-memory-typescript-sdk/issues"},"bin":{"kagura-memory":"dist/cli.js"},"dist":{"shasum":"72083f3ab52c298b3757f004d71ef64d3125474b","tarball":"https://registry.npmjs.org/kagura-memory/-/kagura-memory-0.7.0.tgz","fileCount":11,"integrity":"sha512-Qb9Ms4G1hJevnGen2UTQwh4rspjdKE+VzHe0M2uCnZpNfQlg+a0nRdcO0jCIJexjvVL3BmIHTzzbtXrXVxlLrA==","signatures":[{"sig":"MEUCIQCcvnxP/XZRm9Vu8dO/pSqbAd9dEEOGSTlI2feffbL3KwIgexfel/xYHbnkP6XdHz58v9uTbzsF3/M2jngbvfRltVk=","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"}],"unpackedSize":1498840},"main":"./dist/index.cjs","type":"module","types":"./dist/index.d.ts","module":"./dist/index.js","engines":{"node":">=18"},"exports":{".":{"import":{"types":"./dist/index.d.ts","default":"./dist/index.js"},"require":{"types":"./dist/index.d.cts","default":"./dist/index.cjs"}}},"scripts":{"test":"vitest run","build":"tsup","typecheck":"tsc --noEmit","test:watch":"vitest","prepublishOnly":"npm run typecheck && npm run test && npm run test:no-webcrypto && npm run build","test:no-webcrypto":"vitest run --config vitest.no-webcrypto.config.ts"},"_npmUser":{"name":"kagura-ai","email":"fumikazu.kiyota@gmail.com"},"repository":{"url":"git+https://github.com/kagura-ai/kagura-memory-typescript-sdk.git","type":"git"},"_npmVersion":"10.9.0","description":"TypeScript SDK for Kagura Memory Cloud — memory management, graph exploration, and R2 file storage for AI agents","directories":{},"_nodeVersion":"22.12.0","_hasShrinkwrap":false,"devDependencies":{"tsup":"^8.3.5","vitest":"^2.1.8","typescript":"^5.7.2","@types/node":"^22.10.0","age-encryption":"^0.3.0"},"peerDependencies":{"age-encryption":"^0.3.0"},"peerDependenciesMeta":{"age-encryption":{"optional":true}},"_npmOperationalInternal":{"tmp":"tmp/kagura-memory_0.7.0_1785395545364_0.345958208080434","host":"s3://npm-registry-packages-npm-production"}},"0.8.0":{"name":"kagura-memory","version":"0.8.0","keywords":["ai","memory","mcp","agent","llm","kagura","rag","embeddings","vector-search"],"author":{"name":"Kagura AI","email":"dev@kagura-ai.com"},"license":"MIT","_id":"kagura-memory@0.8.0","maintainers":[{"name":"kagura-ai","email":"fumikazu.kiyota@gmail.com"}],"homepage":"https://github.com/kagura-ai/kagura-memory-typescript-sdk","bugs":{"url":"https://github.com/kagura-ai/kagura-memory-typescript-sdk/issues"},"bin":{"kagura-memory":"dist/cli.js"},"dist":{"shasum":"24bd19b5ce5b803e95c39a9036402352130baf74","tarball":"https://registry.npmjs.org/kagura-memory/-/kagura-memory-0.8.0.tgz","fileCount":11,"integrity":"sha512-GdC0SQW1NFV1Q2BYuFjv7N+Qd7ljwzR18HCu135OqrdYOLObbGnNSz56LTZVxmjcAVuOOP+9A/+5YkZQB7SFJA==","signatures":[{"sig":"MEYCIQCDaVtzVgrjfB96NahDSkjcaz7fd0LrXtFx4GxsbxO6lQIhALwpjkKI2SeLkhcEylUR96JRXiu+jAB69pzK8zlUBBxK","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"}],"attestations":{"url":"https://registry.npmjs.org/-/npm/v1/attestations/kagura-memory@0.8.0","provenance":{"predicateType":"https://slsa.dev/provenance/v1"}},"unpackedSize":2064978},"main":"./dist/index.cjs","type":"module","types":"./dist/index.d.ts","module":"./dist/index.js","engines":{"node":">=18"},"exports":{".":{"import":{"types":"./dist/index.d.ts","default":"./dist/index.js"},"require":{"types":"./dist/index.d.cts","default":"./dist/index.cjs"}}},"gitHead":"61c6f38cfaea10b32f2b5721d722f90a29857c20","scripts":{"test":"vitest run","build":"tsup","typecheck":"tsc --noEmit","test:watch":"vitest","prepublishOnly":"npm run typecheck && npm run test && npm run test:no-webcrypto && npm run build","test:no-webcrypto":"vitest run --config vitest.no-webcrypto.config.ts"},"_npmUser":{"name":"GitHub Actions","email":"npm-oidc-no-reply@github.com","trustedPublisher":{"id":"github","oidcConfigId":"oidc:707d8b6b-f780-4e84-a878-119ad5804768"}},"repository":{"url":"git+https://github.com/kagura-ai/kagura-memory-typescript-sdk.git","type":"git"},"_npmVersion":"11.19.0","description":"TypeScript SDK for Kagura Memory Cloud — memory management, graph exploration, and R2 file storage for AI agents","directories":{},"_nodeVersion":"22.23.1","_hasShrinkwrap":false,"devDependencies":{"tsup":"^8.3.5","vitest":"^2.1.8","typescript":"^5.7.2","@types/node":"^22.10.0","age-encryption":"^0.3.0"},"peerDependencies":{"age-encryption":"^0.3.0"},"peerDependenciesMeta":{"age-encryption":{"optional":true}},"_npmOperationalInternal":{"tmp":"tmp/kagura-memory_0.8.0_1785672939685_0.6346919747159139","host":"s3://npm-registry-packages-npm-production"}},"0.8.1":{"name":"kagura-memory","version":"0.8.1","keywords":["ai","memory","mcp","agent","llm","kagura","rag","embeddings","vector-search"],"author":{"name":"Kagura AI","email":"dev@kagura-ai.com"},"license":"MIT","_id":"kagura-memory@0.8.1","maintainers":[{"name":"kagura-ai","email":"fumikazu.kiyota@gmail.com"}],"homepage":"https://github.com/kagura-ai/kagura-memory-typescript-sdk","bugs":{"url":"https://github.com/kagura-ai/kagura-memory-typescript-sdk/issues"},"bin":{"kagura-memory":"dist/cli.js"},"dist":{"shasum":"d7c587e37c99493357e414f47c7fff082fd1cdf4","tarball":"https://registry.npmjs.org/kagura-memory/-/kagura-memory-0.8.1.tgz","fileCount":11,"integrity":"sha512-XSNTjyBGs9L/sAER9pSsnNQe3c9G4T+wBiNUzEMJSDiB+8Y8q+R8Ze9oHwqdF4eOBnEK0gvSzrKPN5rwL4uldg==","signatures":[{"sig":"MEYCIQDpXzo/hT3SBqxeKNehyZz2iw+fIWhiUXezGlBnRyK7MQIhAISun3rTmVf5dcZpKlCK8HGDEJFDgZ4fXCAHxTjqQmQp","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"},{"sig":"MEYCIQCEZAwENVX8qW6zZ93cQla8fxlWJ9fjwZSQkYuE9xDerAIhAOBxvku46PxO9vqP3u0Nbq33Jx4eD2O+3S4M1yupveij","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"}],"attestations":{"url":"https://registry.npmjs.org/-/npm/v1/attestations/kagura-memory@0.8.1","provenance":{"predicateType":"https://slsa.dev/provenance/v1"}},"unpackedSize":2108410},"main":"./dist/index.cjs","type":"module","types":"./dist/index.d.ts","module":"./dist/index.js","engines":{"node":">=18"},"exports":{".":{"import":{"types":"./dist/index.d.ts","default":"./dist/index.js"},"require":{"types":"./dist/index.d.cts","default":"./dist/index.cjs"}}},"gitHead":"1ab3fc361c7118b5a40582dae86572735904e39f","scripts":{"test":"vitest run","build":"tsup","typecheck":"tsc --noEmit","test:watch":"vitest","prepublishOnly":"npm run typecheck && npm run test && npm run test:no-webcrypto && npm run build","test:no-webcrypto":"vitest run --config vitest.no-webcrypto.config.ts"},"_npmUser":{"name":"GitHub Actions","email":"npm-oidc-no-reply@github.com","trustedPublisher":{"id":"github","oidcConfigId":"oidc:707d8b6b-f780-4e84-a878-119ad5804768"}},"repository":{"url":"git+https://github.com/kagura-ai/kagura-memory-typescript-sdk.git","type":"git"},"_npmVersion":"11.20.0","description":"TypeScript SDK for Kagura Memory Cloud — memory management, graph exploration, and R2 file storage for AI agents","directories":{},"_nodeVersion":"22.23.2","_hasShrinkwrap":false,"devDependencies":{"tsup":"^8.3.5","vitest":"^2.1.8","typescript":"^5.7.2","@types/node":"^22.10.0","age-encryption":"^0.3.0"},"peerDependencies":{"age-encryption":"^0.3.0"},"peerDependenciesMeta":{"age-encryption":{"optional":true}},"_npmOperationalInternal":{"tmp":"tmp/kagura-memory_0.8.1_1790164071137_0.09666958703907724","host":"s3://npm-registry-packages-npm-production"}},"0.9.0":{"name":"kagura-memory","version":"0.9.0","keywords":["ai","memory","mcp","agent","llm","kagura","rag","embeddings","vector-search"],"author":{"name":"Kagura AI","email":"dev@kagura-ai.com"},"license":"MIT","_id":"kagura-memory@0.9.0","maintainers":[{"name":"kagura-ai","email":"fumikazu.kiyota@gmail.com"}],"homepage":"https://github.com/kagura-ai/kagura-memory-typescript-sdk","bugs":{"url":"https://github.com/kagura-ai/kagura-memory-typescript-sdk/issues"},"bin":{"kagura-memory":"dist/cli.js"},"dist":{"shasum":"48a5ea39adbda36f8016409994c2708ca30d4659","tarball":"https://registry.npmjs.org/kagura-memory/-/kagura-memory-0.9.0.tgz","fileCount":11,"integrity":"sha512-1OGqpa+BopE5HsIjqIIRt+r4/AoGT1dxt7WPiKzbLq9IHt+35jDhR2I65BS2pbbIijSo5SDy92UiFys7GVMMsA==","signatures":[{"sig":"MEQCIBU5gM3JGQk6YR3P4Kpwo+v/Faq3rvDE9ywE++DC+r7CAiB+GjHGSYjXhiPzHQCQHGpyo6jj0MNCadSqfsoxWbPUgQ==","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"},{"sig":"MEUCIB/Rtl2/VCB7ptHYIZBaEUe9uDl/CGNAqA68+uwFlkmWAiEAt+IoejhpuHE0DjCw9xEycK0MBYybRHj9YcgrvClF6+E=","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"}],"attestations":{"url":"https://registry.npmjs.org/-/npm/v1/attestations/kagura-memory@0.9.0","provenance":{"predicateType":"https://slsa.dev/provenance/v1"}},"unpackedSize":2386253},"main":"./dist/index.cjs","type":"module","types":"./dist/index.d.ts","module":"./dist/index.js","engines":{"node":">=18"},"exports":{".":{"import":{"types":"./dist/index.d.ts","default":"./dist/index.js"},"require":{"types":"./dist/index.d.cts","default":"./dist/index.cjs"}}},"gitHead":"cde030f2191c3b45cdcf708591f8717744388fd3","scripts":{"test":"vitest run","build":"tsup","typecheck":"tsc --noEmit","test:watch":"vitest","prepublishOnly":"npm run typecheck && npm run test && npm run test:no-webcrypto && npm run build","test:no-webcrypto":"vitest run --config vitest.no-webcrypto.config.ts"},"_npmUser":{"name":"GitHub Actions","email":"npm-oidc-no-reply@github.com","trustedPublisher":{"id":"github","oidcConfigId":"oidc:707d8b6b-f780-4e84-a878-119ad5804768"}},"repository":{"url":"git+https://github.com/kagura-ai/kagura-memory-typescript-sdk.git","type":"git"},"_npmVersion":"11.20.0","description":"TypeScript SDK for Kagura Memory Cloud — memory management, graph exploration, and R2 file storage for AI agents","directories":{},"_nodeVersion":"22.23.2","_hasShrinkwrap":false,"devDependencies":{"tsup":"^8.3.5","vitest":"^2.1.8","typescript":"^5.7.2","@types/node":"^22.10.0","age-encryption":"^0.3.0"},"peerDependencies":{"age-encryption":"^0.3.0"},"peerDependenciesMeta":{"age-encryption":{"optional":true}},"_npmOperationalInternal":{"tmp":"tmp/kagura-memory_0.9.0_1790167704688_0.3432372555881471","host":"s3://npm-registry-packages-npm-production"}},"0.10.0":{"name":"kagura-memory","version":"0.10.0","keywords":["ai","memory","mcp","agent","llm","kagura","rag","embeddings","vector-search"],"author":{"name":"Kagura AI","email":"dev@kagura-ai.com"},"license":"MIT","_id":"kagura-memory@0.10.0","maintainers":[{"name":"kagura-ai","email":"fumikazu.kiyota@gmail.com"}],"homepage":"https://github.com/kagura-ai/kagura-memory-typescript-sdk","bugs":{"url":"https://github.com/kagura-ai/kagura-memory-typescript-sdk/issues"},"bin":{"kagura-memory":"dist/cli.js"},"dist":{"shasum":"51e5b59d2a2464445380f0501bbc25eb243de7af","tarball":"https://registry.npmjs.org/kagura-memory/-/kagura-memory-0.10.0.tgz","fileCount":11,"integrity":"sha512-DDVwOQC1cCsqNfHvlA6ZPFPVI/CHJUUi++izhiSslUM1gji5aRILxGdv+sRZKZO5Xf4QyJxNmpyP3V8gczUAmA==","signatures":[{"sig":"MEUCIAFlLFRfHHbl1RLrtQ2PRBEsrN5g7h8P4TDgXOX/N6PoAiEAnIcUSuAybuuvoKxjlFO2TwYqPAzikUVP9/LXyMcJ0s4=","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"},{"sig":"MEQCIEdm9j+y3EBSSQk8jBIEzUIZMGmAjwaCf1JDgy7FQER4AiB6H07oVinbmQoiqaG42ru1tpbgmMJZnegYY0M/DD0HOg==","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"}],"attestations":{"url":"https://registry.npmjs.org/-/npm/v1/attestations/kagura-memory@0.10.0","provenance":{"predicateType":"https://slsa.dev/provenance/v1"}},"unpackedSize":2616487},"main":"./dist/index.cjs","type":"module","types":"./dist/index.d.ts","module":"./dist/index.js","engines":{"node":">=18"},"exports":{".":{"import":{"types":"./dist/index.d.ts","default":"./dist/index.js"},"require":{"types":"./dist/index.d.cts","default":"./dist/index.cjs"}}},"gitHead":"dddf5e79a818a0939dd69b00bc66990b4a02f041","scripts":{"test":"vitest run","build":"tsup","typecheck":"tsc --noEmit","test:watch":"vitest","prepublishOnly":"npm run typecheck && npm run test && npm run test:no-webcrypto && npm run build","test:no-webcrypto":"vitest run --config vitest.no-webcrypto.config.ts"},"_npmUser":{"name":"GitHub Actions","email":"npm-oidc-no-reply@github.com","trustedPublisher":{"id":"github","oidcConfigId":"oidc:707d8b6b-f780-4e84-a878-119ad5804768"}},"repository":{"url":"git+https://github.com/kagura-ai/kagura-memory-typescript-sdk.git","type":"git"},"_npmVersion":"11.20.0","description":"TypeScript SDK for Kagura Memory Cloud — memory management, graph exploration, and R2 file storage for AI agents","directories":{},"_nodeVersion":"22.23.2","_hasShrinkwrap":false,"devDependencies":{"tsup":"^8.3.5","vitest":"^2.1.8","typescript":"^5.7.2","@types/node":"^22.10.0","age-encryption":"^0.3.0"},"peerDependencies":{"age-encryption":"^0.3.0"},"peerDependenciesMeta":{"age-encryption":{"optional":true}},"_npmOperationalInternal":{"tmp":"tmp/kagura-memory_0.10.0_1790171657857_0.25540236788045423","host":"s3://npm-registry-packages-npm-production"}},"0.10.1":{"name":"kagura-memory","version":"0.10.1","keywords":["ai","memory","mcp","agent","llm","kagura","rag","embeddings","vector-search"],"author":{"name":"Kagura AI","email":"dev@kagura-ai.com"},"license":"MIT","_id":"kagura-memory@0.10.1","maintainers":[{"name":"kagura-ai","email":"fumikazu.kiyota@gmail.com"}],"homepage":"https://github.com/kagura-ai/kagura-memory-typescript-sdk","bugs":{"url":"https://github.com/kagura-ai/kagura-memory-typescript-sdk/issues"},"bin":{"kagura-memory":"dist/cli.js"},"dist":{"shasum":"883c34bd65b8d827376e396e811d7e3078056ecd","tarball":"https://registry.npmjs.org/kagura-memory/-/kagura-memory-0.10.1.tgz","fileCount":11,"integrity":"sha512-zIJE+OJ5JFtdVGQ/d3eEAuQsXbjROdWGSZF6Pi6+RdjBRXxeL1EWNMWjImRlC/fmfwpr7ObjrMRvSpwvpraxrQ==","signatures":[{"sig":"MEUCIQDIJhlXfyaDN0RCwuGvrisaH2CXeve3XIjN4O4h8cI9tQIgUO9K2BUSYaj1oHYcdsbVozAtfRSE1yKd76ZrTyWkDnA=","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"},{"sig":"MEUCIFtN1t81lZlwJJZekxB9ukPT211xDeysQtifMaWOIkyhAiEAsbNiIS9HcXMCFOTPdWVN4VZtr2ikldE9oeFZFJXEik0=","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"}],"attestations":{"url":"https://registry.npmjs.org/-/npm/v1/attestations/kagura-memory@0.10.1","provenance":{"predicateType":"https://slsa.dev/provenance/v1"}},"unpackedSize":2671359},"main":"./dist/index.cjs","type":"module","types":"./dist/index.d.ts","module":"./dist/index.js","engines":{"node":">=18"},"exports":{".":{"import":{"types":"./dist/index.d.ts","default":"./dist/index.js"},"require":{"types":"./dist/index.d.cts","default":"./dist/index.cjs"}}},"gitHead":"29f7b7e21ace1814341fe4465b4f21bba9501a03","scripts":{"test":"vitest run","build":"tsup","typecheck":"tsc --noEmit","test:watch":"vitest","prepublishOnly":"npm run typecheck && npm run test && npm run test:no-webcrypto && npm run build","test:no-webcrypto":"vitest run --config vitest.no-webcrypto.config.ts"},"_npmUser":{"name":"GitHub Actions","email":"npm-oidc-no-reply@github.com","trustedPublisher":{"id":"github","oidcConfigId":"oidc:707d8b6b-f780-4e84-a878-119ad5804768"}},"repository":{"url":"git+https://github.com/kagura-ai/kagura-memory-typescript-sdk.git","type":"git"},"_npmVersion":"11.20.0","description":"TypeScript SDK for Kagura Memory Cloud — memory management, graph exploration, and R2 file storage for AI agents","directories":{},"_nodeVersion":"22.23.2","_hasShrinkwrap":false,"devDependencies":{"tsup":"^8.3.5","vitest":"^2.1.8","typescript":"^5.7.2","@types/node":"^22.10.0","age-encryption":"^0.3.0"},"peerDependencies":{"age-encryption":"^0.3.0"},"peerDependenciesMeta":{"age-encryption":{"optional":true}},"_npmOperationalInternal":{"tmp":"tmp/kagura-memory_0.10.1_1790176003142_0.6303118185974961","host":"s3://npm-registry-packages-npm-production"}},"0.11.0":{"name":"kagura-memory","version":"0.11.0","keywords":["ai","memory","mcp","agent","llm","kagura","rag","embeddings","vector-search"],"author":{"name":"Kagura AI","email":"dev@kagura-ai.com"},"license":"MIT","_id":"kagura-memory@0.11.0","maintainers":[{"name":"kagura-ai","email":"fumikazu.kiyota@gmail.com"}],"homepage":"https://github.com/kagura-ai/kagura-memory-typescript-sdk","bugs":{"url":"https://github.com/kagura-ai/kagura-memory-typescript-sdk/issues"},"bin":{"kagura-memory":"dist/cli.js"},"dist":{"shasum":"db3c86b8e2676c35e1fc0e499774311cdcb51226","tarball":"https://registry.npmjs.org/kagura-memory/-/kagura-memory-0.11.0.tgz","fileCount":11,"integrity":"sha512-cwmqWlRyXOMyGi529x2UJuXh+23cjuCjdo1yHCnChLbVzlDOhf8T4KATDIYNW1GZfv8+sX+QD5n7/i0Z3Tj8GA==","signatures":[{"sig":"MEUCIQDyGQl1taiGcxm+JLVSLweFsuWirVIYgfPx/dnuwoUdtAIgOFZ1W9aENmf1hWG/7pCsp9UeR2a0etE3e4j0znpJS6U=","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"},{"sig":"MEYCIQDbxl/n/dSjlUfM55OX2SE3lEfPfXAZpVQsWc/v21123wIhAJeODU6OW/6qPFg+kV1U3XMnBqgJcEvxWo0RiKLCeA12","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"}],"attestations":{"url":"https://registry.npmjs.org/-/npm/v1/attestations/kagura-memory@0.11.0","provenance":{"predicateType":"https://slsa.dev/provenance/v1"}},"unpackedSize":2784127},"main":"./dist/index.cjs","type":"module","types":"./dist/index.d.ts","module":"./dist/index.js","engines":{"node":">=18"},"exports":{".":{"import":{"types":"./dist/index.d.ts","default":"./dist/index.js"},"require":{"types":"./dist/index.d.cts","default":"./dist/index.cjs"}}},"gitHead":"d331bb2dd4164465d150d62230b6b292a1a37666","scripts":{"test":"vitest run","build":"tsup","typecheck":"tsc --noEmit","test:watch":"vitest","prepublishOnly":"npm run typecheck && npm run test && npm run test:no-webcrypto && npm run build","test:no-webcrypto":"vitest run --config vitest.no-webcrypto.config.ts"},"_npmUser":{"name":"GitHub Actions","email":"npm-oidc-no-reply@github.com","trustedPublisher":{"id":"github","oidcConfigId":"oidc:707d8b6b-f780-4e84-a878-119ad5804768"}},"repository":{"url":"git+https://github.com/kagura-ai/kagura-memory-typescript-sdk.git","type":"git"},"_npmVersion":"11.20.0","description":"TypeScript SDK for Kagura Memory Cloud — memory management, graph exploration, and R2 file storage for AI agents","directories":{},"_nodeVersion":"22.23.2","_hasShrinkwrap":false,"devDependencies":{"tsup":"^8.3.5","vitest":"^2.1.8","typescript":"^5.7.2","@types/node":"^22.10.0","age-encryption":"^0.3.0"},"peerDependencies":{"age-encryption":"^0.3.0"},"peerDependenciesMeta":{"age-encryption":{"optional":true}},"_npmOperationalInternal":{"tmp":"tmp/kagura-memory_0.11.0_1790184366586_0.41434672759723234","host":"s3://npm-registry-packages-npm-production"}},"0.12.0":{"name":"kagura-memory","version":"0.12.0","keywords":["ai","memory","mcp","agent","llm","kagura","rag","embeddings","vector-search"],"author":{"name":"Kagura AI","email":"dev@kagura-ai.com"},"license":"MIT","_id":"kagura-memory@0.12.0","maintainers":[{"name":"kagura-ai","email":"fumikazu.kiyota@gmail.com"}],"homepage":"https://github.com/kagura-ai/kagura-memory-typescript-sdk","bugs":{"url":"https://github.com/kagura-ai/kagura-memory-typescript-sdk/issues"},"bin":{"kagura-memory":"dist/cli.js","kagura-memory-mcp":"dist-mcp/mcp.cjs"},"dist":{"shasum":"ac0ec95e687cfdc32c53b86e61d04e99fce10a00","tarball":"https://registry.npmjs.org/kagura-memory/-/kagura-memory-0.12.0.tgz","fileCount":12,"integrity":"sha512-Gugzc7POjzDkaWzqJLxPYUjZ449reeQIn2MKDcuF+0yWyTSSGBbYDXZTlZmSTwN8xIeuoyZpfofyDVw7h6xXDw==","signatures":[{"sig":"MEQCIA8D13Wc/zGqc0yxbvKajGX1cony3uRKEdgpDDO46OtkAiBpKigJLMD2swReMVNt53iocTSZZl0m9tvhR1XmqjAh/Q==","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"},{"sig":"MEQCIDl//utIGzn4CI85rld4A9WbHY1Pv4Tby+Zs3opI+xn5AiANVhJhk9aTz0CCWPXdMpfn18WMPiJUiKemabh8zABLKQ==","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"}],"attestations":{"url":"https://registry.npmjs.org/-/npm/v1/attestations/kagura-memory@0.12.0","provenance":{"predicateType":"https://slsa.dev/provenance/v1"}},"unpackedSize":3668975},"main":"./dist/index.cjs","type":"module","types":"./dist/index.d.ts","module":"./dist/index.js","engines":{"node":">=18"},"exports":{".":{"import":{"types":"./dist/index.d.ts","default":"./dist/index.js"},"require":{"types":"./dist/index.d.cts","default":"./dist/index.cjs"}}},"gitHead":"e6581eff148b1d21be4526e4c0ea369bd9556220","scripts":{"test":"vitest run","build":"tsup","test:mcpb":"node scripts/smoke-mcpb.mjs","typecheck":"tsc --noEmit","build:mcpb":"npm run build && node scripts/build-mcpb.mjs","test:watch":"vitest","prepublishOnly":"npm run typecheck && npm run test && npm run test:no-webcrypto && npm run build","test:no-webcrypto":"vitest run --config vitest.no-webcrypto.config.ts"},"_npmUser":{"name":"GitHub Actions","email":"npm-oidc-no-reply@github.com","trustedPublisher":{"id":"github","oidcConfigId":"oidc:707d8b6b-f780-4e84-a878-119ad5804768"}},"repository":{"url":"git+https://github.com/kagura-ai/kagura-memory-typescript-sdk.git","type":"git"},"_npmVersion":"11.20.0","description":"TypeScript SDK for Kagura Memory Cloud — memory management, graph exploration, and R2 file storage for AI agents","directories":{},"_nodeVersion":"22.23.2","_hasShrinkwrap":false,"devDependencies":{"tsup":"^8.3.5","fflate":"0.8.3","vitest":"^2.1.8","typescript":"^5.7.2","@types/node":"^22.10.0","age-encryption":"^0.3.0","@anthropic-ai/mcpb":"2.1.2"},"peerDependencies":{"age-encryption":"^0.3.0"},"peerDependenciesMeta":{"age-encryption":{"optional":true}},"_npmOperationalInternal":{"tmp":"tmp/kagura-memory_0.12.0_1790301968135_0.5136931431426579","host":"s3://npm-registry-packages-npm-production"}},"0.13.0":{"name":"kagura-memory","version":"0.13.0","keywords":["ai","memory","mcp","agent","llm","kagura","rag","embeddings","vector-search"],"author":{"name":"Kagura AI","email":"dev@kagura-ai.com"},"license":"MIT","_id":"kagura-memory@0.13.0","maintainers":[{"name":"kagura-ai","email":"fumikazu.kiyota@gmail.com"}],"homepage":"https://github.com/kagura-ai/kagura-memory-typescript-sdk","bugs":{"url":"https://github.com/kagura-ai/kagura-memory-typescript-sdk/issues"},"bin":{"kagura-memory":"dist/cli.js","kagura-memory-mcp":"dist-mcp/mcp.cjs"},"dist":{"shasum":"5b29df52c11ab1f4b4e96c3358ca031ea31569d6","tarball":"https://registry.npmjs.org/kagura-memory/-/kagura-memory-0.13.0.tgz","fileCount":12,"integrity":"sha512-NxUMXo1cWlful/L6WjI+jkaK0L9GZ/ye1cm95YddGxMvz2AI+IR40x9RaV4v0dadiDS9mkBwI88x6hxzcEHx3A==","signatures":[{"sig":"MEYCIQDKQiGWSSatjFNlpih7y7hoxaYATgugt6wbWwQCcO2P3AIhAIa6eZqQsO17G3xUsy3kdnudXApCBNivfJDC8CfZdKqM","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"},{"sig":"MEUCICb6sI//eJC92fZLySwVTGhCWvspHJhB6ajEucGo2iT4AiEAjeJlI/IpnGKQaTwsXt3isALXX3f+PSsBFJfVH4ycV1w=","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"}],"attestations":{"url":"https://registry.npmjs.org/-/npm/v1/attestations/kagura-memory@0.13.0","provenance":{"predicateType":"https://slsa.dev/provenance/v1"}},"unpackedSize":3920426},"main":"./dist/index.cjs","type":"module","types":"./dist/index.d.ts","module":"./dist/index.js","engines":{"node":">=18"},"exports":{".":{"import":{"types":"./dist/index.d.ts","default":"./dist/index.js"},"require":{"types":"./dist/index.d.cts","default":"./dist/index.cjs"}}},"gitHead":"2cf4cc8fc2cbc175fa181ef1ee11be12f67557c8","scripts":{"test":"vitest run","build":"tsup","test:mcpb":"node scripts/smoke-mcpb.mjs","typecheck":"tsc --noEmit","build:mcpb":"npm run build && node scripts/build-mcpb.mjs","test:watch":"vitest","prepublishOnly":"npm run typecheck && npm run test && npm run test:no-webcrypto && npm run build","test:no-webcrypto":"vitest run --config vitest.no-webcrypto.config.ts"},"_npmUser":{"name":"GitHub Actions","email":"npm-oidc-no-reply@github.com","trustedPublisher":{"id":"github","oidcConfigId":"oidc:707d8b6b-f780-4e84-a878-119ad5804768"}},"repository":{"url":"git+https://github.com/kagura-ai/kagura-memory-typescript-sdk.git","type":"git"},"_npmVersion":"11.20.0","description":"TypeScript SDK for Kagura Memory Cloud — memory management, graph exploration, and R2 file storage for AI agents","directories":{},"_nodeVersion":"22.23.2","_hasShrinkwrap":false,"devDependencies":{"tsup":"^8.3.5","fflate":"0.8.3","vitest":"^2.1.8","typescript":"^5.7.2","@types/node":"^22.10.0","age-encryption":"^0.3.0","@anthropic-ai/mcpb":"2.1.2"},"peerDependencies":{"age-encryption":"^0.3.0"},"peerDependenciesMeta":{"age-encryption":{"optional":true}},"_npmOperationalInternal":{"tmp":"tmp/kagura-memory_0.13.0_1790335236998_0.69799581076052","host":"s3://npm-registry-packages-npm-production"}},"0.14.0":{"name":"kagura-memory","version":"0.14.0","keywords":["ai","memory","mcp","agent","llm","kagura","rag","embeddings","vector-search"],"author":{"name":"Kagura AI","email":"dev@kagura-ai.com"},"license":"MIT","_id":"kagura-memory@0.14.0","maintainers":[{"name":"kagura-ai","email":"fumikazu.kiyota@gmail.com"}],"homepage":"https://github.com/kagura-ai/kagura-memory-typescript-sdk","bugs":{"url":"https://github.com/kagura-ai/kagura-memory-typescript-sdk/issues"},"bin":{"kagura-memory":"dist/cli.js","kagura-memory-mcp":"dist-mcp/mcp.cjs"},"dist":{"shasum":"3d5a2d848d56513576ec912a7bbeca0c2ed7af16","tarball":"https://registry.npmjs.org/kagura-memory/-/kagura-memory-0.14.0.tgz","fileCount":12,"integrity":"sha512-Ta9n1ePE5litHdta8h0OHWjosr5E2epntFQwhX5bCN0xW7oR5tjYnEEokIIRp0PJOCpmeC9zzu5DHjTLaGum3Q==","signatures":[{"sig":"MEUCIQCBE/Cg2TClAqZqIBWrD2CEOCmAMWS23hbqu8pY0PQspQIgZJAWFe9OEYqPQciSUwzqkEqeejCFhKGHggFUgMQwjFU=","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"},{"sig":"MEYCIQCS5rUB7vbxH/9fjzm2y99ANFnduigl3Laqq59Ky35yDwIhALHrLH1Q0R2BdKQ13uHzBm1ibhMiltXNXtzEM2fcxJs4","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"}],"attestations":{"url":"https://registry.npmjs.org/-/npm/v1/attestations/kagura-memory@0.14.0","provenance":{"predicateType":"https://slsa.dev/provenance/v1"}},"unpackedSize":4354349},"main":"./dist/index.cjs","type":"module","types":"./dist/index.d.ts","module":"./dist/index.js","engines":{"node":">=18"},"exports":{".":{"import":{"types":"./dist/index.d.ts","default":"./dist/index.js"},"require":{"types":"./dist/index.d.cts","default":"./dist/index.cjs"}}},"gitHead":"6d92ec36b6a01cfa620d3405e7ac886ef3634b47","scripts":{"test":"vitest run","build":"tsup","test:mcpb":"node scripts/smoke-mcpb.mjs","typecheck":"tsc --noEmit","build:mcpb":"npm run build && node scripts/build-mcpb.mjs","test:watch":"vitest","prepublishOnly":"npm run typecheck && npm run test && npm run test:no-webcrypto && npm run build","test:no-webcrypto":"vitest run --config vitest.no-webcrypto.config.ts"},"_npmUser":{"name":"GitHub Actions","email":"npm-oidc-no-reply@github.com","trustedPublisher":{"id":"github","oidcConfigId":"oidc:707d8b6b-f780-4e84-a878-119ad5804768"}},"repository":{"url":"git+https://github.com/kagura-ai/kagura-memory-typescript-sdk.git","type":"git"},"_npmVersion":"11.20.0","description":"TypeScript SDK for Kagura Memory Cloud — memory management, graph exploration, and R2 file storage for AI agents","directories":{},"_nodeVersion":"22.23.2","_hasShrinkwrap":false,"devDependencies":{"tsup":"^8.3.5","fflate":"0.8.3","vitest":"^2.1.8","typescript":"^5.7.2","@types/node":"^22.10.0","age-encryption":"^0.3.0","@anthropic-ai/mcpb":"2.1.2"},"peerDependencies":{"age-encryption":"^0.3.0"},"peerDependenciesMeta":{"age-encryption":{"optional":true}},"_npmOperationalInternal":{"tmp":"tmp/kagura-memory_0.14.0_1790522215005_0.6699214602470576","host":"s3://npm-registry-packages-npm-production"}},"0.14.1":{"name":"kagura-memory","version":"0.14.1","keywords":["ai","memory","mcp","agent","llm","kagura","rag","embeddings","vector-search"],"author":{"name":"Kagura AI","email":"dev@kagura-ai.com"},"license":"MIT","_id":"kagura-memory@0.14.1","maintainers":[{"name":"kagura-ai","email":"fumikazu.kiyota@gmail.com"}],"homepage":"https://github.com/kagura-ai/kagura-memory-typescript-sdk","bugs":{"url":"https://github.com/kagura-ai/kagura-memory-typescript-sdk/issues"},"bin":{"kagura-memory":"dist/cli.js","kagura-memory-mcp":"dist-mcp/mcp.cjs"},"dist":{"shasum":"68993e1a36ef3d7558c86c4756e1ed7e01013136","tarball":"https://registry.npmjs.org/kagura-memory/-/kagura-memory-0.14.1.tgz","fileCount":12,"integrity":"sha512-VUhnScPgSBj0ED6R4VCLO99ETDwSCIxGVYwE+KupELS79wOgfAgylXjkT/x2HETHOJlAyyGozaKOJ7NLqN+MqQ==","signatures":[{"sig":"MEUCIAFrMVRhtOHkGY5hcPaSQOFrK22NgQnLYzVGtZucBt1YAiEAkBrOUjcnRBsf5NPvzZH4K7DU9qRnFwpRpafackJoIi4=","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"},{"sig":"MEYCIQDdcZgAtBuLjBBbLh1kwjCFbCALZuUoivTs0WmkJB4raAIhAM0U+P3c61GlE6+LVTce0ZvOOSPByv+U0p0qznMXuA2/","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"}],"attestations":{"url":"https://registry.npmjs.org/-/npm/v1/attestations/kagura-memory@0.14.1","provenance":{"predicateType":"https://slsa.dev/provenance/v1"}},"unpackedSize":4371002},"main":"./dist/index.cjs","type":"module","types":"./dist/index.d.ts","module":"./dist/index.js","engines":{"node":">=18"},"exports":{".":{"import":{"types":"./dist/index.d.ts","default":"./dist/index.js"},"require":{"types":"./dist/index.d.cts","default":"./dist/index.cjs"}}},"gitHead":"4061ecc05c0ef97af4be23403b18aa7a024cc07f","scripts":{"test":"vitest run","build":"tsup","test:mcpb":"node scripts/smoke-mcpb.mjs","typecheck":"tsc --noEmit","build:mcpb":"npm run build && node scripts/build-mcpb.mjs","test:watch":"vitest","prepublishOnly":"npm run typecheck && npm run test && npm run test:no-webcrypto && npm run build","test:no-webcrypto":"vitest run --config vitest.no-webcrypto.config.ts"},"_npmUser":{"name":"GitHub Actions","email":"npm-oidc-no-reply@github.com","trustedPublisher":{"id":"github","oidcConfigId":"707d8b6b-f780-4e84-a878-119ad5804768"}},"overrides":{"tsup":{"esbuild":"^0.28.1"},"external-editor":{"tmp":"^0.2.6"}},"repository":{"url":"git+https://github.com/kagura-ai/kagura-memory-typescript-sdk.git","type":"git"},"_npmVersion":"11.21.0","description":"TypeScript SDK for Kagura Memory Cloud — memory management, graph exploration, and R2 file storage for AI agents","directories":{},"_nodeVersion":"22.23.3","_hasShrinkwrap":false,"devDependencies":{"tsup":"^8.3.5","vite":"^6.4.3","fflate":"0.8.3","vitest":"^3.2.7","typescript":"^5.7.2","@types/node":"^22.10.0","age-encryption":"^0.3.0","@anthropic-ai/mcpb":"2.1.2"},"peerDependencies":{"age-encryption":"^0.3.0"},"peerDependenciesMeta":{"age-encryption":{"optional":true}},"_npmOperationalInternal":{"tmp":"tmp/kagura-memory_0.14.1_1790936956957_0.5413790672389218","host":"s3://npm-registry-packages-npm-production"}},"0.15.0":{"name":"kagura-memory","version":"0.15.0","keywords":["ai","memory","mcp","agent","llm","kagura","rag","embeddings","vector-search"],"author":{"name":"Kagura AI","email":"dev@kagura-ai.com"},"license":"MIT","_id":"kagura-memory@0.15.0","maintainers":[{"name":"kagura-ai","email":"fumikazu.kiyota@gmail.com"}],"homepage":"https://github.com/kagura-ai/kagura-memory-typescript-sdk","bugs":{"url":"https://github.com/kagura-ai/kagura-memory-typescript-sdk/issues"},"bin":{"kagura-memory":"dist/cli.js","kagura-memory-mcp":"dist-mcp/mcp.cjs"},"dist":{"shasum":"26b4bd3a2ff48c99bdd664218e728b4047fec595","tarball":"https://registry.npmjs.org/kagura-memory/-/kagura-memory-0.15.0.tgz","fileCount":12,"integrity":"sha512-opA5TopWiYD9qpCwZkKdrvhmA50aD0V0Jpt4FyLAroRSCRzk857zMHFqGHc25D8xppXRYaUaCW49Dk6x6vGQIQ==","signatures":[{"sig":"MEUCIQD44/zirmdONPdsvwYF+TKiKc40V25c2ehTemb2cKkESQIgY37jDtwJEVD7tIQILjLPh6eewwharsuD8BSNYZOJspk=","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"},{"sig":"MEUCIDdA3xgl/czwZ7HkV2PsqNDsJhgt95/2W+i/F5qCL7q/AiEAyM6DqHZVLxdRNNW30o2cHFIgs8986Fx8ujy0TovBTTE=","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"}],"attestations":{"url":"https://registry.npmjs.org/-/npm/v1/attestations/kagura-memory@0.15.0","provenance":{"predicateType":"https://slsa.dev/provenance/v1"}},"unpackedSize":6032526},"main":"./dist/index.cjs","type":"module","types":"./dist/index.d.ts","module":"./dist/index.js","engines":{"node":">=20.19.0"},"exports":{".":{"import":{"types":"./dist/index.d.ts","default":"./dist/index.js"},"require":{"types":"./dist/index.d.cts","default":"./dist/index.cjs"}}},"gitHead":"3fa67471586bd5a4144a3b1166c2d63df4ddcd70","scripts":{"test":"vitest run","build":"tsup","test:mcpb":"node scripts/smoke-mcpb.mjs","typecheck":"tsc --noEmit","build:mcpb":"npm run build && node scripts/build-mcpb.mjs","test:watch":"vitest","prepublishOnly":"npm run typecheck && npm run test && npm run build"},"_npmUser":{"name":"GitHub Actions","email":"npm-oidc-no-reply@github.com","trustedPublisher":{"id":"github","oidcConfigId":"707d8b6b-f780-4e84-a878-119ad5804768"}},"overrides":{"tsup":{"esbuild":"^0.28.1"},"external-editor":{"tmp":"^0.2.6"}},"repository":{"url":"git+https://github.com/kagura-ai/kagura-memory-typescript-sdk.git","type":"git"},"_npmVersion":"11.21.1","description":"TypeScript SDK for Kagura Memory Cloud — memory management, graph exploration, and R2 file storage for AI agents","directories":{},"_nodeVersion":"22.23.4","_hasShrinkwrap":false,"devDependencies":{"tsup":"^8.3.5","fflate":"0.8.3","vitest":"^4.1.12","typescript":"^5.7.2","@types/node":"^22.10.0","age-encryption":"^0.3.0","@anthropic-ai/mcpb":"2.1.2"},"peerDependencies":{"age-encryption":"^0.3.0"},"peerDependenciesMeta":{"age-encryption":{"optional":true}},"_npmOperationalInternal":{"tmp":"tmp/kagura-memory_0.15.0_1793815285481_0.26382194365841827","host":"s3://npm-registry-packages-npm-production"}},"0.16.0":{"name":"kagura-memory","version":"0.16.0","keywords":["ai","memory","mcp","agent","llm","kagura","rag","embeddings","vector-search"],"author":{"name":"Kagura AI","email":"dev@kagura-ai.com"},"license":"MIT","_id":"kagura-memory@0.16.0","maintainers":[{"name":"kagura-ai","email":"fumikazu.kiyota@gmail.com"}],"homepage":"https://github.com/kagura-ai/kagura-memory-typescript-sdk","bugs":{"url":"https://github.com/kagura-ai/kagura-memory-typescript-sdk/issues"},"bin":{"kagura-memory":"dist/cli.js","kagura-memory-mcp":"dist-mcp/mcp.cjs"},"dist":{"shasum":"c7bab8092dca4e13b3106c11b29dae32aa0c4d02","tarball":"https://registry.npmjs.org/kagura-memory/-/kagura-memory-0.16.0.tgz","fileCount":12,"integrity":"sha512-HPcRuEmITANv2jFITbazlB/uXYXB6OXtGgyVxBHUI+fupxQlnM4AV3im2B6RlreN3sxU4FSM0rNsz8n7ayk9gg==","signatures":[{"sig":"MEUCIARSJm65BTHPo0QioH2lw7syNyehmUVwCSTKI+0Lfb3HAiEAnFJIZUixisQ7BM+zBcM0uHor2/gC9gb0s+ZQ42EIenk=","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"},{"sig":"MEQCIBkPV2Lxp69DtUTYohyEpD+HuyZbFDWhgusQWXu4uqatAiBp0M4USmqS62xznT3WHADNGv80f6bPsaAmClKWJtbqZg==","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"}],"attestations":{"url":"https://registry.npmjs.org/-/npm/v1/attestations/kagura-memory@0.16.0","provenance":{"predicateType":"https://slsa.dev/provenance/v1"}},"unpackedSize":6108868},"main":"./dist/index.cjs","type":"module","types":"./dist/index.d.ts","module":"./dist/index.js","engines":{"node":">=20.19.0"},"exports":{".":{"import":{"types":"./dist/index.d.ts","default":"./dist/index.js"},"require":{"types":"./dist/index.d.cts","default":"./dist/index.cjs"}}},"gitHead":"bb462079137fc78938b86f03847e57b38e9a0929","scripts":{"test":"vitest run","build":"tsup","test:mcpb":"node scripts/smoke-mcpb.mjs","typecheck":"tsc --noEmit","build:mcpb":"npm run build && node scripts/build-mcpb.mjs","test:watch":"vitest","prepublishOnly":"npm run typecheck && npm run test && npm run build"},"_npmUser":{"name":"GitHub Actions","email":"npm-oidc-no-reply@github.com","trustedPublisher":{"id":"github","oidcConfigId":"707d8b6b-f780-4e84-a878-119ad5804768"}},"overrides":{"tsup":{"esbuild":"^0.28.1"},"external-editor":{"tmp":"^0.2.6"}},"repository":{"url":"git+https://github.com/kagura-ai/kagura-memory-typescript-sdk.git","type":"git"},"_npmVersion":"11.21.2","description":"TypeScript SDK for Kagura Memory Cloud — memory management, graph exploration, and R2 file storage for AI agents","directories":{},"_nodeVersion":"22.23.4","_hasShrinkwrap":false,"devDependencies":{"tsup":"^8.3.5","fflate":"0.8.3","vitest":"^4.1.12","typescript":"^5.7.2","@types/node":"^22.10.0","age-encryption":"^0.3.0","@anthropic-ai/mcpb":"2.1.2"},"peerDependencies":{"age-encryption":"^0.3.0"},"peerDependenciesMeta":{"age-encryption":{"optional":true}},"_npmOperationalInternal":{"tmp":"tmp/kagura-memory_0.16.0_1793848801909_0.20901376616327838","host":"s3://npm-registry-packages-npm-production"}},"0.17.0":{"name":"kagura-memory","version":"0.17.0","keywords":["ai","memory","mcp","agent","llm","kagura","rag","embeddings","vector-search"],"author":{"name":"Kagura AI","email":"dev@kagura-ai.com"},"license":"MIT","_id":"kagura-memory@0.17.0","maintainers":[{"name":"kagura-ai","email":"fumikazu.kiyota@gmail.com"}],"homepage":"https://github.com/kagura-ai/kagura-memory-typescript-sdk","bugs":{"url":"https://github.com/kagura-ai/kagura-memory-typescript-sdk/issues"},"bin":{"kagura-memory":"dist/cli.js","kagura-memory-mcp":"dist-mcp/mcp.cjs"},"dist":{"shasum":"9431a8509765449cc259b2687f172ea03aba4794","tarball":"https://registry.npmjs.org/kagura-memory/-/kagura-memory-0.17.0.tgz","fileCount":12,"integrity":"sha512-7f2xjeT5Y034P7UUbXtEjPLieHiZ8/XN3TA6WduFNEORJqp8KHdN1A6vNJcEmwQUMGnqWb0eV0ICtHFe3WUIqw==","signatures":[{"sig":"MEQCIA+SBzO74j9cw77uTNqO1nLWTZCxHeddLZCn/pH7YmCaAiB7du4WlgsnaUPD9n/fyV07zUFYcGlPnTJp+jcpk4P3hQ==","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"},{"sig":"MEUCIQDMU0eIWZRJ1bNfGylmD9gnW9lCXD24eVWQfupwe/mvdwIgHqVDTE/hAWcoHnaO6dCpUH+5L8T8pSIL5cRs6bend3E=","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"}],"attestations":{"url":"https://registry.npmjs.org/-/npm/v1/attestations/kagura-memory@0.17.0","provenance":{"predicateType":"https://slsa.dev/provenance/v1"}},"unpackedSize":6182558},"main":"./dist/index.cjs","type":"module","types":"./dist/index.d.ts","module":"./dist/index.js","engines":{"node":">=22.12.0"},"exports":{".":{"import":{"types":"./dist/index.d.ts","default":"./dist/index.js"},"require":{"types":"./dist/index.d.cts","default":"./dist/index.cjs"}}},"gitHead":"4c47c18e21fa04ab4e873e196e7a930d0a201896","scripts":{"test":"vitest run","build":"tsup","test:mcpb":"node scripts/smoke-mcpb.mjs","typecheck":"tsc --noEmit","build:mcpb":"npm run build && node scripts/build-mcpb.mjs","test:watch":"vitest","prepublishOnly":"npm run typecheck && npm run test && npm run build"},"_npmUser":{"name":"GitHub Actions","email":"npm-oidc-no-reply@github.com","trustedPublisher":{"id":"github","oidcConfigId":"707d8b6b-f780-4e84-a878-119ad5804768"}},"overrides":{"tsup":{"esbuild":"^0.28.1"},"external-editor":{"tmp":"^0.2.6"}},"repository":{"url":"git+https://github.com/kagura-ai/kagura-memory-typescript-sdk.git","type":"git"},"_npmVersion":"11.21.2","description":"TypeScript SDK for Kagura Memory Cloud — memory management, graph exploration, and R2 file storage for AI agents","directories":{},"_nodeVersion":"22.23.4","_hasShrinkwrap":false,"devDependencies":{"tsup":"^8.3.5","fflate":"0.8.3","vitest":"^5.1.0","typescript":"^5.7.2","@types/node":"^22.10.0","age-encryption":"^0.3.0","@anthropic-ai/mcpb":"2.1.2"},"peerDependencies":{"age-encryption":"^0.3.0"},"peerDependenciesMeta":{"age-encryption":{"optional":true}},"_npmOperationalInternal":{"tmp":"tmp/kagura-memory_0.17.0_1793851239167_0.6918625766262377","host":"s3://npm-registry-packages-npm-production"}},"0.18.0":{"name":"kagura-memory","version":"0.18.0","keywords":["ai","memory","mcp","agent","llm","kagura","rag","embeddings","vector-search"],"author":{"name":"Kagura AI","email":"dev@kagura-ai.com"},"license":"MIT","_id":"kagura-memory@0.18.0","maintainers":[{"name":"kagura-ai","email":"fumikazu.kiyota@gmail.com"}],"homepage":"https://github.com/kagura-ai/kagura-memory-typescript-sdk","bugs":{"url":"https://github.com/kagura-ai/kagura-memory-typescript-sdk/issues"},"bin":{"kagura-memory":"dist/cli.js","kagura-memory-mcp":"dist-mcp/mcp.cjs"},"dist":{"shasum":"7242b644330a01c7e5376fe7715a38309e22b357","tarball":"https://registry.npmjs.org/kagura-memory/-/kagura-memory-0.18.0.tgz","fileCount":12,"integrity":"sha512-eNREdFz2boi6lcmN3PL/jb7ffk1NGlcnnvYPRV+kqZzsCWrz5udE1HoxlOEwSNJmECsI3snEK/OPF0i3+Ep/WA==","signatures":[{"sig":"MEUCIDCltXNv5NTPQ41d/g0YC2kpc43a7cHx2yxcAXfXnCIYAiEAzu+wFgQ6vMtO+V6x9NwiP5BqRiJabsQ9NjrOTCNLaM0=","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"},{"sig":"MEYCIQDZ0pIc7LmTHkswWGgajVUHvdvcx8wFJEo6vXj4UOFvOgIhAMBvHRf7uOJMwKeAcRYzML33UsqMoJwn7u4FdptTNSyy","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"}],"attestations":{"url":"https://registry.npmjs.org/-/npm/v1/attestations/kagura-memory@0.18.0","provenance":{"predicateType":"https://slsa.dev/provenance/v1"}},"unpackedSize":6350654},"main":"./dist/index.cjs","type":"module","types":"./dist/index.d.ts","module":"./dist/index.js","engines":{"node":">=22.12.0"},"exports":{".":{"import":{"types":"./dist/index.d.ts","default":"./dist/index.js"},"require":{"types":"./dist/index.d.cts","default":"./dist/index.cjs"}}},"gitHead":"ef1741375bcf23d6d8ef8dc3255bd70dc669dbb6","scripts":{"test":"vitest run","build":"tsup","test:mcpb":"node scripts/smoke-mcpb.mjs","typecheck":"tsc --noEmit","build:mcpb":"npm run build && node scripts/build-mcpb.mjs","test:watch":"vitest","prepublishOnly":"npm run typecheck && npm run test && npm run build"},"_npmUser":{"name":"GitHub Actions","email":"npm-oidc-no-reply@github.com","trustedPublisher":{"id":"github","oidcConfigId":"707d8b6b-f780-4e84-a878-119ad5804768"}},"overrides":{"tsup":{"esbuild":"^0.28.1"},"external-editor":{"tmp":"^0.2.6"}},"repository":{"url":"git+https://github.com/kagura-ai/kagura-memory-typescript-sdk.git","type":"git"},"_npmVersion":"11.21.2","description":"TypeScript SDK for Kagura Memory Cloud — memory management, graph exploration, and R2 file storage for AI agents","directories":{},"_nodeVersion":"22.23.4","_hasShrinkwrap":false,"devDependencies":{"tsup":"^8.3.5","fflate":"0.8.3","vitest":"^5.1.0","typescript":"^5.7.2","@types/node":"^22.10.0","age-encryption":"^0.3.0","@anthropic-ai/mcpb":"2.1.2"},"peerDependencies":{"age-encryption":"^0.3.0"},"peerDependenciesMeta":{"age-encryption":{"optional":true}},"_npmOperationalInternal":{"tmp":"tmp/kagura-memory_0.18.0_1794300558147_0.12281549862141805","host":"s3://npm-registry-packages-npm-production"}},"0.18.1":{"name":"kagura-memory","version":"0.18.1","keywords":["ai","memory","mcp","agent","llm","kagura","rag","embeddings","vector-search"],"author":{"name":"Kagura AI","email":"dev@kagura-ai.com"},"license":"MIT","_id":"kagura-memory@0.18.1","maintainers":[{"name":"kagura-ai","email":"fumikazu.kiyota@gmail.com"}],"homepage":"https://github.com/kagura-ai/kagura-memory-typescript-sdk","bugs":{"url":"https://github.com/kagura-ai/kagura-memory-typescript-sdk/issues"},"bin":{"kagura-memory":"dist/cli.js","kagura-memory-mcp":"dist-mcp/mcp.cjs"},"dist":{"shasum":"789e2c5131fc66fbebdd5930367510f0174b8617","tarball":"https://registry.npmjs.org/kagura-memory/-/kagura-memory-0.18.1.tgz","fileCount":12,"integrity":"sha512-aP0+48iB1M4QisIHI/0xYUlBeRgNg46JkFoarH2PAR+OsUWX+DKkRZQl5vKtmaZYh2SYVoMAdYkIN1Mdeqn8uA==","signatures":[{"sig":"MEYCIQDDEsT9GofJMaMXnDPN7AgItMvUGkebhc4r5JJNP87NcgIhANHS4VRXIqoGBbmftOCdW16c2GCllbbkjGs5V9F/bxtM","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"},{"sig":"MEUCICT3yZqoqlRS4TZECk/7pUgHELQ9CL2kZnW4xvMtjbGEAiEA1AHNVqtEDVt5H7hwonclWZjy2zANiApnJ+bw76kmrek=","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"}],"attestations":{"url":"https://registry.npmjs.org/-/npm/v1/attestations/kagura-memory@0.18.1","provenance":{"predicateType":"https://slsa.dev/provenance/v1"}},"unpackedSize":6356252},"main":"./dist/index.cjs","type":"module","types":"./dist/index.d.ts","module":"./dist/index.js","engines":{"node":">=22.12.0"},"exports":{".":{"import":{"types":"./dist/index.d.ts","default":"./dist/index.js"},"require":{"types":"./dist/index.d.cts","default":"./dist/index.cjs"}}},"gitHead":"71f6de9c1c181917a2294bd67410dead569ced62","scripts":{"test":"vitest run","build":"tsup","test:mcpb":"node scripts/smoke-mcpb.mjs","typecheck":"tsc --noEmit","build:mcpb":"npm run build && node scripts/build-mcpb.mjs","test:watch":"vitest","prepublishOnly":"npm run typecheck && npm run test && npm run build"},"_npmUser":{"name":"GitHub Actions","email":"npm-oidc-no-reply@github.com","trustedPublisher":{"id":"github","oidcConfigId":"707d8b6b-f780-4e84-a878-119ad5804768"}},"overrides":{"tsup":{"esbuild":"^0.28.1"},"external-editor":{"tmp":"^0.2.6"}},"repository":{"url":"git+https://github.com/kagura-ai/kagura-memory-typescript-sdk.git","type":"git"},"_npmVersion":"11.21.2","description":"TypeScript SDK for Kagura Memory Cloud — memory management, graph exploration, and R2 file storage for AI agents","directories":{},"_nodeVersion":"22.23.4","_hasShrinkwrap":false,"devDependencies":{"tsup":"^8.3.5","fflate":"0.8.3","vitest":"^5.1.0","typescript":"^5.7.2","@types/node":"^22.10.0","age-encryption":"^0.3.0","@anthropic-ai/mcpb":"2.1.2"},"peerDependencies":{"age-encryption":"^0.3.0"},"peerDependenciesMeta":{"age-encryption":{"optional":true}},"_npmOperationalInternal":{"tmp":"tmp/kagura-memory_0.18.1_1794302693285_0.5944159807140176","host":"s3://npm-registry-packages-npm-production"}},"0.19.0":{"name":"kagura-memory","version":"0.19.0","keywords":["ai","memory","mcp","agent","llm","kagura","rag","embeddings","vector-search"],"author":{"name":"Kagura AI","email":"dev@kagura-ai.com"},"license":"MIT","_id":"kagura-memory@0.19.0","maintainers":[{"name":"kagura-ai","email":"fumikazu.kiyota@gmail.com"}],"homepage":"https://github.com/kagura-ai/kagura-memory-typescript-sdk","bugs":{"url":"https://github.com/kagura-ai/kagura-memory-typescript-sdk/issues"},"bin":{"kagura-memory":"dist/cli.js","kagura-memory-mcp":"dist-mcp/mcp.cjs"},"dist":{"shasum":"05b3c31473c752abb77ef57fa86008778e77125f","tarball":"https://registry.npmjs.org/kagura-memory/-/kagura-memory-0.19.0.tgz","fileCount":12,"integrity":"sha512-Zv81+sIK2B06hnYyqrgxttpM0xr0Prqlung2PwRrG3NSkrrpJGhUiSHLKWpjVLHPofNwhfN5s0sRzCnKImAHjg==","signatures":[{"sig":"MEYCIQDL0VE4jGptjbEH38bnDvvmusBH7DyXevcdAI2tzhqtVgIhAN4ANgvgv/UryEoLyTehYfaFwaNykkAtjoieLbpkLwwl","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"},{"sig":"MEQCICPj0y70KobhUmBTLZRrEV5JZPMe6LBNYa4cSirDpZXnAiBPT424LhGQmtR4YuRIRlEc/2Mf6W4l4Dr65BXzsXsA1g==","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"}],"attestations":{"url":"https://registry.npmjs.org/-/npm/v1/attestations/kagura-memory@0.19.0","provenance":{"predicateType":"https://slsa.dev/provenance/v1"}},"unpackedSize":6501263},"main":"./dist/index.cjs","type":"module","types":"./dist/index.d.ts","module":"./dist/index.js","engines":{"node":">=22.12.0"},"exports":{".":{"import":{"types":"./dist/index.d.ts","default":"./dist/index.js"},"require":{"types":"./dist/index.d.cts","default":"./dist/index.cjs"}}},"gitHead":"bb0ad3bec91ab02d50876c2ac50f7df898faad41","scripts":{"test":"vitest run","build":"tsup","test:mcpb":"node scripts/smoke-mcpb.mjs","typecheck":"tsc --noEmit","build:mcpb":"npm run build && node scripts/build-mcpb.mjs","test:watch":"vitest","prepublishOnly":"npm run typecheck && npm run test && npm run build"},"_npmUser":{"name":"GitHub Actions","email":"npm-oidc-no-reply@github.com","trustedPublisher":{"id":"github","oidcConfigId":"707d8b6b-f780-4e84-a878-119ad5804768"}},"overrides":{"tsup":{"esbuild":"^0.28.1"},"external-editor":{"tmp":"^0.2.6"}},"repository":{"url":"git+https://github.com/kagura-ai/kagura-memory-typescript-sdk.git","type":"git"},"_npmVersion":"11.21.2","description":"TypeScript SDK for Kagura Memory Cloud — memory management, graph exploration, and file storage for AI agents","directories":{},"_nodeVersion":"22.23.4","_hasShrinkwrap":false,"devDependencies":{"tsup":"^8.3.5","fflate":"0.8.3","vitest":"^5.1.0","typescript":"^5.7.2","@types/node":"^22.10.0","age-encryption":"^0.3.0","@anthropic-ai/mcpb":"2.1.2"},"peerDependencies":{"age-encryption":"^0.3.0"},"peerDependenciesMeta":{"age-encryption":{"optional":true}},"_npmOperationalInternal":{"tmp":"tmp/kagura-memory_0.19.0_1794616154464_0.6974955252864133","host":"s3://npm-registry-packages-npm-production"}},"0.19.1":{"_id":"kagura-memory@0.19.1","bin":{"kagura-memory":"dist/cli.js","kagura-memory-mcp":"dist-mcp/mcp.cjs"},"bugs":{"url":"https://github.com/kagura-ai/kagura-memory-typescript-sdk/issues"},"dist":{"shasum":"b1f2f9a5fc4d6b1f0e01cd511e18d08353f469c2","tarball":"https://registry.npmjs.org/kagura-memory/-/kagura-memory-0.19.1.tgz","fileCount":12,"integrity":"sha512-36uc7om75O9GUo3K5AjQGy+AiGBWmuqCouYnDRUdSkjAoypkqF+6yQ7G0XUhovIEVYaezVrC2dzsQX8PTSqmBQ==","signatures":[{"sig":"MEQCIBBu4VkywEotDjNZ5LOzmUrLYtwntBMjzj/Svd7VXmydAiA43ivkU20EmImhAh6cuhM7VlabEBiYYRM8GXl6Itpi3A==","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"},{"keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U","sig":"MEUCIDaFNESaDzieU/u1vdAXqa+ibpTYQ27xRqV2asHOt624AiEArK6vJRIlpXrxjzpZBp5yteIYhj2X9B49lA/FffXvcyo="}],"attestations":{"url":"https://registry.npmjs.org/-/npm/v1/attestations/kagura-memory@0.19.1","provenance":{"predicateType":"https://slsa.dev/provenance/v1"}},"unpackedSize":6506887},"main":"./dist/index.cjs","name":"kagura-memory","type":"module","types":"./dist/index.d.ts","author":{"name":"Kagura AI","email":"dev@kagura-ai.com"},"module":"./dist/index.js","engines":{"node":">=22.12.0"},"exports":{".":{"import":{"types":"./dist/index.d.ts","default":"./dist/index.js"},"require":{"types":"./dist/index.d.cts","default":"./dist/index.cjs"}}},"gitHead":"1c7d44a272743ee45aba1e50ef389a0bc7e8fb9d","license":"MIT","scripts":{"test":"vitest run","build":"tsup","test:mcpb":"node scripts/smoke-mcpb.mjs","typecheck":"tsc --noEmit","build:mcpb":"npm run build && node scripts/build-mcpb.mjs","test:watch":"vitest","prepublishOnly":"npm run typecheck && npm run test && npm run build"},"version":"0.19.1","_npmUser":{"name":"GitHub Actions","email":"npm-oidc-no-reply@github.com","trustedPublisher":{"id":"github","oidcConfigId":"707d8b6b-f780-4e84-a878-119ad5804768"}},"homepage":"https://github.com/kagura-ai/kagura-memory-typescript-sdk","keywords":["ai","memory","mcp","agent","llm","kagura","rag","embeddings","vector-search"],"overrides":{"tsup":{"esbuild":"^0.28.1"},"external-editor":{"tmp":"^0.2.6"}},"repository":{"url":"git+https://github.com/kagura-ai/kagura-memory-typescript-sdk.git","type":"git"},"_npmVersion":"11.21.3","description":"TypeScript SDK for Kagura Memory Cloud — memory management, graph exploration, and file storage for AI agents","directories":{},"maintainers":[{"name":"kagura-ai","email":"fumikazu.kiyota@gmail.com"}],"_nodeVersion":"22.23.4","_hasShrinkwrap":false,"devDependencies":{"tsup":"^8.3.5","fflate":"0.8.3","vitest":"^5.1.0","typescript":"^5.7.2","@types/node":"^22.10.0","age-encryption":"^0.3.0","@anthropic-ai/mcpb":"2.1.2"},"peerDependencies":{"age-encryption":"^0.3.0"},"peerDependenciesMeta":{"age-encryption":{"optional":true}},"_npmOperationalInternal":{"host":"s3://npm-registry-packages-npm-production","tmp":"tmp/kagura-memory_0.19.1_1794983720938_0.529665373036758"}}},"time":{"created":"2026-07-06T08:04:37.483Z","modified":"2026-11-18T06:35:21.592Z","0.1.0":"2026-07-06T08:04:37.709Z","0.2.0":"2026-07-16T05:39:01.160Z","0.3.0":"2026-07-16T06:27:42.681Z","0.5.0":"2026-07-28T12:19:45.009Z","0.6.0":"2026-07-28T14:47:17.582Z","0.7.0":"2026-07-30T07:12:25.588Z","0.8.0":"2026-08-02T12:15:39.899Z","0.8.1":"2026-09-23T11:47:51.243Z","0.9.0":"2026-09-23T12:48:24.780Z","0.10.0":"2026-09-23T13:54:18.013Z","0.10.1":"2026-09-23T15:06:43.261Z","0.11.0":"2026-09-23T17:26:06.704Z","0.12.0":"2026-09-25T02:06:08.511Z","0.13.0":"2026-09-25T11:20:37.142Z","0.14.0":"2026-09-27T15:16:55.112Z","0.14.1":"2026-10-02T10:29:17.091Z","0.15.0":"2026-11-04T18:01:25.610Z","0.16.0":"2026-11-05T03:20:02.023Z","0.17.0":"2026-11-05T04:00:39.321Z","0.18.0":"2026-11-10T08:49:18.260Z","0.18.1":"2026-11-10T09:24:53.421Z","0.19.0":"2026-11-14T00:29:14.608Z","0.19.1":"2026-11-18T06:35:21.111Z"},"bugs":{"url":"https://github.com/kagura-ai/kagura-memory-typescript-sdk/issues"},"author":{"name":"Kagura AI","email":"dev@kagura-ai.com"},"license":"MIT","homepage":"https://github.com/kagura-ai/kagura-memory-typescript-sdk","keywords":["ai","memory","mcp","agent","llm","kagura","rag","embeddings","vector-search"],"repository":{"url":"git+https://github.com/kagura-ai/kagura-memory-typescript-sdk.git","type":"git"},"description":"TypeScript SDK for Kagura Memory Cloud — memory management, graph exploration, and file storage for AI agents","maintainers":[{"name":"kagura-ai","email":"fumikazu.kiyota@gmail.com"}],"readme":"<p align=\"center\">\n  <strong>Kagura Memory SDK for TypeScript</strong> — Node.js client for <a href=\"https://github.com/kagura-ai/memory-cloud\">Kagura Memory Cloud</a>\n</p>\n\n## What is this?\n\nThis SDK connects your TypeScript/JavaScript code to [Kagura Memory Cloud](https://github.com/kagura-ai/memory-cloud), giving AI agents the ability to **remember, search, and learn** from past interactions. It is a TypeScript port of the [Python SDK](https://github.com/kagura-ai/kagura-memory-python-sdk) core:\n\n| Client | Protocol | Use Case |\n|--------|----------|----------|\n| **`KaguraClient`** | MCP (JSON-RPC) | Direct memory ops — remember, recall, explore, reference, forget |\n| **`ResourceClient`** | REST API | External data ingestion — push data from Slack, CI/CD, CRM into Kagura |\n| **`FilesClient`** | REST + presigned PUT | File uploads with sha256 integrity binding |\n| **`WorkspaceClient`** | REST API | Workspace member, invitation, and API key management; operation history (web session) |\n| **`SecretClient`** | REST API | [Zero-knowledge secrets](#zero-knowledge-secrets) — age-encrypted on your machine, opaque to the server |\n| **`AgentsClient`** | REST API | Agent bootstrap for API-key-only callers (no MCP session) |\n| **`MemoryClient`** | REST API | [Tool guardrails](#tool-guardrails-over-rest-memory-cloud-v0740) for API-key-only callers — the `AGENTS.md` export digest and the guardrail set |\n\nA `kagura-memory` command-line tool ships alongside it, mirroring the Python\nCLI's `kagura` command — see [Command line](#command-line).\n\nFor Claude Desktop, the [MCPB extension](docs/mcpb.md) provides browser sign-in\nand a refresh-aware connection without Python or a global CLI install. The npm\npackage also includes `kagura-memory-mcp`, a transparent stdio/HTTP proxy.\n\n## Installation\n\n```bash\nnpm install kagura-memory\n```\n\nRequires Node.js >= 22.12.0 (this runtime migration drops Node 20). Zero\nruntime dependencies — the one optional peer dependency, for\n[zero-knowledge secrets](#the-crypto-package-is-opt-in), is never installed\nunless you ask for it.\n\nTargets memory-cloud **v0.75.0** (`MIN_SERVER_VERSION`) and is checked\nagainst memory-cloud up to **v0.170.0**; the Python SDK 0.44.0, its\nparity target, is checked up to v0.142.0 (v0.143.0 changes nothing a\nclient sees), and its own `MIN_SERVER_VERSION` stays at an advisory\n0.17.1. `checkServerVersion()` warns, and never throws, on an older\nserver, which still answers: it reports a tool it predates as not found\nand leaves out fields it predates, and an option it predates is ignored\nbefore v0.82.0 and refused (`invalid_argument`) from v0.82.0. A\npre-release of the minimum (`0.75.0-rc1`) counts as older, and a version\nit cannot read (`0.75`, `main-abc123`) gets no warning, as in the Python\nSDK. The method notes below say which server version a feature needs, and\n[`docs/compatibility.md`](docs/compatibility.md) lists every floor by\nrelease.\n\n## Quick Start\n\n```ts\nimport { KaguraClient } from \"kagura-memory\";\n\n// Credentials resolve automatically: explicit apiKey > KAGURA_API_KEY env\n// > OAuth profile (~/.kagura/credentials.json, shared with the Python CLI)\n// > .kagura.json\nconst client = new KaguraClient();\n\n// Store a memory\nconst stored = await client.remember({\n  contextId: \"your-context-id\",\n  summary: \"User prefers TypeScript strict mode\",\n  content: \"Enabled strict + noUncheckedIndexedAccess in all new projects.\",\n  type: \"note\",\n  tags: [\"typescript\", \"preferences\"],\n});\n\n// Search memories (hybrid semantic + keyword)\nconst hits = await client.recall({\n  contextId: \"your-context-id\",\n  query: \"typescript preferences\",\n  k: 5,\n});\n\nfor (const memory of hits.results as Array<Record<string, unknown>>) {\n  console.log(memory.summary);\n}\n\nawait client.close();\n```\n\n### Authentication\n\nFour ways to authenticate, in resolution order:\n\n1. **Explicit key** — `new KaguraClient({ apiKey: \"kagura_...\" })`\n2. **Environment** — `KAGURA_API_KEY` (+ optional `KAGURA_MCP_URL`)\n3. **OAuth profile** — `~/.kagura/credentials.json`, written by `login()`\n   (below) or the Python CLI's `kagura auth login`. Profiles\n   (`KAGURA_PROFILE` env or `{ profile: \"name\" }`) and auto-refresh work\n   exactly like the Python SDK; the credentials file is shared between both\n   SDKs. Every refresh (a client's, `kagura-memory-mcp`'s, `refresh()`'s,\n   `auth refresh`'s and `auth token`'s) takes one lock on the file and\n   re-reads it first, so processes sharing a profile refresh it once, and a\n   `credentials.json` that exists but cannot be read is never rewritten (see\n   [Refreshing](#refreshing)).\n4. **Config file** — `.kagura.json` in the working directory or home:\n\n```json\n{\n  \"api_key\": \"kagura_your_api_key\",\n  \"mcp_url\": \"https://memory.kagura-ai.com/mcp\"\n}\n```\n\nCredentials travel over HTTPS only. Every client, `login()`, `auth login\n--server` and the URL every `setup` subcommand puts in a harness entry\n(`setup claude`, `setup codex`, `setup hermes` and `setup openclaw`)\nrefuse a plain-HTTP URL unless its host is `localhost`, `127.0.0.1` or\n`[::1]`; `doctor` warns about a resolved MCP URL like that and skips the\nserver check, as the Python CLI does, and fails a configured `mcp_url`\nlike that which the credential does not use (see `doctor`). The URL is\nread as `fetch` reads it, so the scheme in any case (`HTTP://`), without\nits slashes (`http:host`), or with whitespace or control characters\naround it or a tab or newline inside it is refused all the same, as in\nthe Python SDK.\n\n### Command line\n\n`kagura-memory` mirrors the Python CLI's `kagura` command: its subcommands\nand flags take the Python CLI's names, and it prints the same JSON on\nstdout and exits with the same codes (2 for a usage error, 1 for a runtime\nfailure). An option a command, a group or the root does not take is\nrefused in the words of click 8.3, the version the Python CLI's lockfile\npins: `Error: No such option: --x`, naming the option without any value\ngiven in it, and for a long option close to one the command takes,\nclick's suggestion after it (`Did you mean --json?`, or `(Possible\noptions: --end, --period)` for more than one); click 8.4 and later write\n`No such option '--x'.`. An option given no value is `Error: Option '-w'\nrequires an argument.`. Either error wins over `--help`, as click reads\nthe whole command line first, and only the first one is reported,\nfollowed by the command's help (where click prints its usage line and a\n`Try '… --help'` hint first). An unknown option does not win over\n`--help` where the Python CLI passes such options on\n(`ignore_unknown_options`: `secret exec`, and `measure record` for a\nnegative VALUE), so `secret exec --help --bogus` prints the help; without\n`--help`, `secret exec` still refuses an unknown option before the child\ncommand rather than running it as that command. The one command not\nported, the few options only this bin has, and the deliberate\ndifferences are listed below. The parity target is the Python CLI\n**0.44.0**.\n\nValues are read as click reads them. A choice matches exactly, as click's\n`Choice` does, except where the Python CLI declares one case-insensitive\n(`--source-type`, `--progress`), casefolded as click folds it; a number\nis read as Python's `int()` and `float()` read it, `1_000`, any decimal\ndigit (`١٢`, `１２`) and the whitespace they skip included (a BOM is no\nwhitespace to them); an id argument (`TOKEN_ID`, `INVITATION_ID`,\n`KEY_ID`) is sent exactly, an integer however large, or a string id\n(`rtok_…`, `winv_…`, `akey_…`, memory-cloud v0.89.0+) as typed; and an\nout-of-range value is echoed as click converted it (`files upload\n--importance 2` is refused as `2.0 is not in the range 0.0<=x<=1.0.`).\nA failure prints `Error: <message>` on stderr, and a quota or plan\nrefusal adds the lines the Python CLI adds: `  Resets at: <time>`, in\nPython's `isoformat()` form, and `  Required plan: <label> (<key>)`.\n\n```bash\nnpx kagura-memory --help\n```\n\n| Group | Commands |\n|---|---|\n| `auth` | `login` `logout` `refresh` `status` `use` `list` `token` `create-key` `list-keys` `revoke-key` |\n| `context` | `list` `create` `update` `delete` `search-config` (plus the `contexts` alias) |\n| memory | `remember` `recall` `reference` `forget` `update-memory` `explore` `bootstrap` `changes` `list` |\n| `edge` | `list` `create` `update` `delete` |\n| `sleep` | `history` `report` `rollback` |\n| `measure` | `record` `series` |\n| `files` | `upload` `list` `delete` `download-url` |\n| `guardrails` | `load` `digest` |\n| `resource` | `tokens {list,create,update,revoke}` `list` `setup` `schema` `stats` `indexer-status` `events` `ingest` `ingest-batch` `import` |\n| `secret` | `keygen` `list` `put` `get` `grant` `revoke` `rotate` `delete` `pubkeys` `approve` `audit-verify` `exec` |\n| `setup` | `claude` `codex` `hermes` `openclaw` |\n| `workspace` | `member {list,add,set-role,remove}` `invite {create,list,revoke}` |\n| other | `config show` `doctor` |\n\n```bash\nnpx kagura-memory auth login --profile work --read-only\nnpx kagura-memory recall \"OAuth setup\" -c dev -k 10\nnpx kagura-memory recall \"dependency injection\" --no-rerank   # skip reranking; no flag follows the context config (v0.69.0+)\nnpx kagura-memory recall \"project context\" --verbosity full   # uncut summaries + tags; compact is the server default (v0.108.0+)\nnpx kagura-memory recall \"deploy runbook\" --full 1 --max-chars 40000   # content on the top hit only (v0.114.0+); reply budget (v0.82.0+)\n# A server that refuses one of --verbosity/--full/--max-chars (invalid_argument) gets one\n# retry without the ones it names (every one it sent when it names none), with a note on stderr.\nnpx kagura-memory remember -s \"FastAPI DI\" --content \"Use Depends()\" --tags \"python,fastapi\"\nnpx kagura-memory remember -s \"Release 2.0 shipped\" --content \"...\" --observed-at 2026-11-01   # when the fact held (v0.127.0+)\nnpx kagura-memory remember -s \"Deploy key rotated\" --content \"...\" --dedupe check   # a duplicate_candidate prints with exit 0 (v0.93.0+)\nnpx kagura-memory update-memory -m MEM_UUID --add-tag status:done --remove-tag status:open \\\n  --details-patch '{\"status\": \"done\"}' --expected-version 3   # partial edits (v0.140.0+), guarded (v0.113.0+)\nnpx kagura-memory update-memory -m MEM_UUID --accept-supersede-candidate   # shadow the suggested older memory (v0.126.0+)\nnpx kagura-memory context search-config CTX_UUID --content-chunks --no-llm-selector   # owner only (v0.110.0+ / v0.115.0+)\nnpx kagura-memory bootstrap --since 7d --format index --max-chars 5000   # open a session (v0.93.0+; index/--max-chars < 10000 v0.115.0+)\nnpx kagura-memory changes 2026-11-04T09:00:00Z --by-me   # what you wrote since then (v0.93.0+; --by-me v0.114.0+)\nnpx kagura-memory list --filters '{\"type\": \"task\", \"details.status\": \"open\"}' --order-by created_at   # exact filters, a total, no ranking\nnpx kagura-memory list --filters '{\"event_after\": \"2026-10-01\", \"event_before\": \"2026-11-01\"}' --order-by event_start   # by the event's own date (v0.152.0+)\nnpx kagura-memory forget --ids \"$ID1,$ID2\"   # one row per id: deleted / skipped / error (v0.117.0+)\nnpx kagura-memory measure record <context-id> weight_kg 71.5 --unit kg\nnpx kagura-memory setup codex --dry-run   # show the command and the entry; change nothing\nnpx kagura-memory setup claude --add-tools explore --context <context-id>   # ?add_tools= (v0.129.0+), ?context= (v0.114.0+)\nnpx kagura-memory guardrails digest <context-id> --target instructions --add-tools explore  # …for ?add_tools=explore\nnpx kagura-memory doctor   # ends with the tools/list view of the CLI's MCP URL\n```\n\nThe context id comes from `-c/--context-id`, or from `context_id` in\n`.kagura.json`. Credentials live in `~/.kagura/credentials.json` and are\nshared with the Python CLI, so either tool can create a profile the other\nthen uses.\n\n**`bootstrap`, `changes`, `list` and `forget --ids`** (memory-cloud\nv0.93.0+; `forget --ids` v0.117.0+) are the Python CLI 0.44.0's session\ncommands. Each prints the Python SDK model's dump of the reply: a field\nthe server left out prints with its default, a timestamp in pydantic's\nform. `list` is the MCP `list` tool (`listMatching`) and takes\n`--filters` as a JSON object; it is no longer a spelling of `auth list`,\nthe one bare `auth` alias that gave way. `forget --ids` takes 1-50\ncomma-separated memory UUIDs and refuses `-k` (exit 1, as Python); when\nevery id failed it prints one `<id> <status> [<error>]` row per id on\nstderr before the error, so the `not_attempted` ones can be resent.\n\n**`context list`** prints the `list_contexts` reply as JSON on stdout.\nFrom memory-cloud v0.82.0 that is one page: 100 contexts by default, or\n`--limit N` (1-200), and a page also stops at 20,000 characters. When a\npage was cut, a line on stderr says so, as the Python CLI's does:\n`has_more: true; next_cursor: \"100\". For the next page, run the command\nagain with the same options and --cursor \"100\".` The cursor is printed\nthere only when it is ASCII digits, the only form the server takes back;\nany other `next_cursor` is left to the JSON (`--cursor set to next_cursor\nfrom the JSON`), so nothing a shell would expand or a terminal would\ninterpret reaches stderr. A blank `--cursor` is unset, so a script can feed\nback `jq -r '.next_cursor // empty'`. `--details` with a `--limit` above 20\nand no non-blank `--name-contains` is a usage error (exit 2) before any\nrequest, since the server refuses that page. Servers before v0.82.0 return\nevery context and ignore `--limit` and `--cursor`. **`contexts`**, the\nshort form, takes no options and prints one page, the 100 most recently\nused contexts; when more follow, its stderr line names the\n`kagura-memory context list --cursor \"100\"` call for the next page.\n\n```bash\nnpx kagura-memory context list --limit 50\nnpx kagura-memory context list --limit 50 --cursor 50   # the next_cursor of the page before\n```\n\n**`update-memory --merge-details`** (and `--details-patch`, a patch you\nwrite yourself) sends `--details` / `--location` as one `details_patch`\ncall, as the Python CLI 0.44.0 does: the server merges their top-level keys\nover the stored details, keeps the rest, and deletes a key given as `null`.\nBefore memory-cloud 0.140.0 there is no `details_patch`, so the command\nfalls back to reading the memory with `reference()` and writing the merged\nobject — two calls, not one atomic update — and says so on stderr:\n\n| Server | First call | Then | Exit |\n|---|---|---|---|\n| 0.140.0+ | `update_memory` with `details_patch` and every other flag | nothing | 0 |\n| 0.82.0–0.139.x | refused: `invalid_argument` listing exactly `details_patch` in `unknown_arguments`, nothing written | `reference()`, then the merged object as `details` with every other flag (`--expected-version` still guards it on 0.113.0+) | 0 |\n| 0.82.0+, refused for any other reason (or naming another argument too) | the error | nothing | 1 |\n| before 0.82.0 | ignored: success without `version`, the other flags applied (or ignored) | `GET /api/v1/system/info`; below 0.82.0, `reference()` and the merged details alone | 0, or 1 naming each of `--add-tag` / `--remove-tag` / `--expected-version` / `--accept-supersede-candidate` / `--observed-at` it sent, after printing the reply |\n| a reply without `version` from 0.82.0+, an unparseable version or a failed probe | — | nothing more written | 1 |\n\nThe fallback merge follows the server's rule (`null` deletes) and folds\n`--observed-at` into the merged details. A bounded `reference()` reply\n(`details_omitted`, a `details_json` page) is refused, never merged.\n`--details-patch` cannot be combined with `--details` / `--location` /\n`--merge-details`, `--tags` replaces the whole list where `--add-tag` /\n`--remove-tag` edit it, and a refused `--expected-version` prints the\nserver's message with `  Current version: N` (and `  Help:`).\n\n**`resource setup`** names the new context after the resource, or `-n/--name`\n(lowercase letters, digits, `-` and `_`, at most 100), which a workspace\nthat already has a context of that name needs. It prints Python's six\nfields, `context_id`, `context_name`, `resource_id`, `token`, `token_id`\nand `warning` (null when the server sends none); the token is shown this\nonce. `--summary` is accepted and ignored, with a note on stderr, since\nthe server's `setup_resource` has no summary: the context's owner sets it\nafterwards with `kagura-memory context update <context_id> --summary …`.\nAs in Python, `resource tokens revoke` prints `Token revoked.`, and\n`resource schema` prints `No schema registered for this resource.` when\nthere is none (the route's 404).\n\n**`resource list`** (`ResourceClient.listResources()`) also lists a resource\nwhose contexts were all deleted while it still holds active tokens, with\n`context_id` and `context_name` null (memory-cloud v0.94.0+), so those\ntokens stay reachable; revoking its last token removes the row.\n\n**Records print as the Python CLI prints them.** `resource tokens\nlist|create|update`, `resource list|stats|indexer-status|events|schema`,\n`resource ingest|ingest-batch|setup`, `files list` and `files upload`\nprint the result as the Python CLI does, `model_dump_json(indent=2)` of\nthe Python SDK's model\n([#66](https://github.com/kagura-ai/kagura-memory-typescript-sdk/issues/66)):\nthe model's keys in its order, its default for a key the server left out\n(`null`, `false`, `0`, `[]`, `{}`, `\"success\"`), and no key it does not\nhave. Each value is what the model's type makes of it: `\"12\"` for an int\nprints `12`, a float prints as one (`1.0`, `1e-6`), a resource event's id\npast 2^53 prints exactly, and a timestamp prints in pydantic's form\n(`2026-06-01T09:00:00.5+00:00` as `2026-06-01T09:00:00.500000Z`), which\nis the form memory-cloud already sends. A record the model refuses (a\nrequired field missing, a value of the wrong type) exits 1 with the\nPython SDK's `KaguraResponseError` text, e.g. `ResourceClient.list_tokens:\nunexpected server response for PaginatedResourceTokensResponse\n(tokens.0.created_at: Field required). …`, and so does a body that is no\nrecord at all (`null`). An untyped mapping nested past pydantic's limit\n(256 containers) exits 1 with pydantic's `Error serializing to JSON:\nValueError: Circular reference detected (depth exceeded)`. The body is\nread as Python's `json.loads` reads it\n([#69](https://github.com/kagura-ai/kagura-memory-typescript-sdk/issues/69)),\nso an int field sent `9007199254740993` prints exactly, one sent `1e20`\nexits 1 as pydantic refuses it (`Unable to parse input string as an\ninteger, exceeded maximum size`), and a float field sent `-0` prints\n`0.0`. Inside an untyped mapping (an event's `payload`, a batch's\n`errors`) the keys print in the server's order (`{\"b\": 1, \"2\": 2}` stays\nso) and each number as Python read it (`1.0`, `1e+16`, `-0.0`, an\ninteger past 2^53 exactly). `NaN` and `Infinity` are read, and print\n`null` as pydantic writes them; an integer of more than 4,300 digits\nmakes the body non-JSON, as Python's `int()` limit does; and a body\nnested deeper than 973 containers exits 1 with Python's `maximum\nrecursion depth exceeded while decoding a JSON array from a unicode\nstring`. A string value holding a lone surrogate (`\"\\ud800\"`) exits 1\nwith pydantic's `Error serializing to JSON: UnicodeEncodeError: 'utf-8'\ncodec can't encode character '\\ud800' in position 0: surrogates not\nallowed`; a key of the untyped mapping itself holding one prints as\nthree U+FFFD, as pydantic converts it, and a key of a mapping nested\ninside it exits 1 like a value. Three differences remain. Python's depth\nis its recursion limit less the stack in use: 973 is what these commands\nreach, and `doctor` given such a `/api/v1/system/info` body reports a\nfailed server check where Python stops with a traceback. The body is\ndecoded as UTF-8, as `fetch` decodes every body: one that is not valid\nUTF-8 is read with U+FFFD in place of the bad bytes, where Python calls\nit non-JSON, or, for a lone surrogate sent as its three raw bytes\n(`ED A0 80`), reads it and then refuses the dump; and a UTF-16 or UTF-32\nbody, with or without a BOM, which Python's `json.loads` detects and\nreads, is non-JSON here (`Server returned a non-JSON body (HTTP 200)\n…`). No memory-cloud server sends either. And only REST bodies are read\nthis way: the MCP commands (`recall`, `remember`, …) still print a whole\nfloat as `1` and list an object's integer keys first.\n\n**Files and imports.** The `files` commands take their workspace from the\ncredential's own source, as the Python CLI does (its #115): `-c` when\ngiven, else the OAuth profile's workspace, else the `context_id` of the\n`.kagura.json` that holds the key; a `KAGURA_API_KEY` key needs `-c`.\n`files delete` prints `Deleted <file_id>` and `files download-url` the bare\nURL. `files upload --remember` writes the memory the Python CLI writes\n(`source_uri` the file's `file://` URI, `details` the file's id, sha256,\nsize and type), and prints `{\"file\": …, \"memory\": …}`, the file as its\nmodel reads it. `resource import` detects the format by the file's\nextension, so stdin needs `--format`, and reads CSV and JSON the way\nPython's `csv` and `json` modules read them, with their error messages\n(exit 1). A row keeps its keys in the order read, and a doc_id taken from\na number is Python's `str()` of it as written (`1234567890123456789`,\n`10.0`, `1e+20`). A `NaN` or `Infinity` in the input fails the batch\nholding it, as the Python CLI's request encoding does (`Out of range float\nvalues are not JSON compliant: nan`), rather than being sent as `null`. A\ndoc_id outside 1-255 characters and a CSV row with more cells than the\nheader are refused before anything is sent, where the Python CLI stops\nwith a traceback. Two limits of JavaScript numbers remain: a number in the\npayload is sent as JavaScript reads it, so an integer past 2^53 loses its\nlast digits and `10.0` is sent as `10`, where Python sends them as\nwritten; and `-V` above 2^53 is refused as too large (exit 2), where\nPython sends it. The summary prints as Python's `json.dumps` prints it\n([#69](https://github.com/kagura-ai/kagura-memory-typescript-sdk/issues/69)):\n`created` and `failed` add up exactly past 2^53, and each error keeps\nthe server's key order and numbers (`1.0`, `1e+16`, `NaN`); a lone\nsurrogate in an error's string or key prints as `?` (one per code unit,\n`\"😀???x\"`), as the Python CLI's UTF-8 stdout with `errors=\"replace\"`\nwrites what `json.dumps` kept. The counts in\n`--progress json`'s events (`detail.created_so_far`, the success event's\n`detail.created`) are still JavaScript numbers, so past 2^53 they are\nrounded where Python's are exact; `-v` prints no counts.\n\n**Progress.** `files upload` and `resource import` take `-v/--verbose`\n(repeatable) and `--progress rich|json|none`. Progress goes to stderr, one\nline per event, so stdout carries the same result either way.\n`--progress json` writes the Python CLI's NDJSON,\n`{\"v\": 1, \"ts\": …, \"stage\": …, \"kind\": …, \"msg\": …, \"detail\": …}`, whose\nlast event is the operation's single `success` or `error`; `-v`, or\n`--progress rich`, writes the lines the Python CLI shows\n(`→ Reserving upload report.pdf (1234 bytes)`, `✓ Upload complete`,\n`✗ Upload failed: …`); `--progress none` is silent even with `-v`.\nAs in the Python CLI (0.41.1+), `files upload --remember` reports success\nonly once the linked memory is written (a failed write ends the stream\nwith that error), and `resource import` starts its stream only once its\nclient is built, so a credential that fails prints the error alone. A\nstderr closed early (`2>&1 | head -1`) does not stop the upload or the\nimport. Ctrl-C ends the process without a final event (Node's default\nSIGINT handling exits at once), where the Python CLI catches the\ninterrupt and ends the stream with `error`.\n\n**Connecting a harness.** Each `setup` subcommand sets up an MCP entry\nnamed `kagura-memory`: the URL plus a Bearer header, or, with\n`--url-form --oauth` (below), the URL alone, which the harness signs in\nto itself. The key is never printed, and no harness CLI gets it on its\ncommand line.\n\n`setup claude` also writes `.kagura.json` (0600, gitignored). Its key\ncomes from `--api-key`, else the `api_key` in the project's own\n`.kagura.json` (the one in `--project-dir`), else `KAGURA_API_KEY`; the\nURL likewise from `--mcp-url`, else that file's `mcp_url`, else\n`KAGURA_MCP_URL`, an empty `--mcp-url` counting as none, as Python's `or`\nreads it. It never takes them from `~/.kagura.json`, nor, with\n`--project-dir`, from the current directory's `.kagura.json`: those are\nanother project's credentials. `.kagura.json` gets the URL as given; the\nparameters that `--guardrails`, `--tool-profile`, `--add-tools` and\n`--context` set go on the entry's URL only. A plain-HTTP URL to a host other than localhost is refused,\nwhichever of the three it came from, before anything is written or run\n(exit 1, `Connection failed: MCP URL must use HTTPS for security …`, the\nPython CLI's words, whose connection test refuses it).\n\n`setup claude` makes no connection unless the context (`--context-id`,\nelse that file's `context_id`, else `KAGURA_CONTEXT_ID`) is a name, which\nit resolves through `list_contexts` on that key and URL, as the Python\nCLI 0.44.0 does (python-sdk#309): any value but a UUID in the dashed form\n(a dashless 32-hex one included) up to 100 characters is trimmed and\nmatched by id or exact name on the first page, then by exact name on\nevery page, with `name_contains` and the cursor when the first page says\n`has_more` (memory-cloud 0.73.0 / 0.82.0; an older server's one reply is\nthe whole list), at most 50 pages, and the context's id is written. A\nvalue no page matches, or a longer one, is still written, trimmed, after\n`Warning: no context named 'NAME' is visible to this credential; using it\nas a raw context id.` on stderr when it is not a UUID, although the\nserver takes only a UUID there, so later commands refuse it. Pages that\ndo not add up (`Setup failed: …`) and a failed request (`Authentication\nfailed: …` or `Cannot connect to …` with Python's hints, else\n`Connection failed: …` on the first page) stop setup (exit 1) before\nanything is written or run, as does a blank key, which the lookup never\nreplaces with another credential, and a context that is only whitespace\n(`Invalid context_id: '' — must be alphanumeric, hyphens, or underscores\nonly`, Python's words); an empty `--context-id` counts as none, as in\nPython. A dashed UUID is never looked up, even\none that is a context's name, which Python 0.44.0 would resolve when that\ncontext is on its first page.\n\nWithout `--agents-md`, `setup codex`, `setup hermes` and `setup openclaw`\nnever see, write, print or pass the key, as in the Python CLI. The entry\nnames the environment variable the harness reads the key from, and the\nnotes say where to put it; a missing key is no error. These three write\nno config file themselves: no `.kagura.json`, no `.gitignore` line, no\nharness `.env`. The harness's own CLI writes the entry, or you add the\nprinted block. The one file they may write is the AGENTS.md export\n(`--agents-md`, below), which is fetched with the usual credential: with\nit, a missing credential stops setup (exit 1).\n\n| Subcommand | How the entry is applied | Where the key goes |\n|---|---|---|\n| `setup claude` | `.mcp.json` (`--scope project`, the default), or `claude mcp add-json --scope user …` | `--scope project`: in `.mcp.json`; `--scope user`: `KAGURA_MCP_API_KEY`, exported where Claude Code starts |\n| `setup codex` | `codex mcp add … --bearer-token-env-var KAGURA_API_KEY` (with `--oauth`: `codex mcp add NAME --url URL`, attached, when there is a terminal and no `-y`) | `export KAGURA_API_KEY=…` in the shell profile that starts Codex |\n| `setup hermes` | the `config.yaml` block is printed; `hermes mcp add` is interactive, and this port never prompts (with `--oauth`: `hermes mcp add … --auth oauth`, attached, when there is a terminal and no `-y`) | `MCP_KAGURA_MEMORY_API_KEY=…` (`MCP_<NAME>_API_KEY` with `--name`) in the `.env` beside `config.yaml`, added with an editor |\n| `setup openclaw` | `openclaw mcp add … --transport streamable-http --no-probe`, or `openclaw mcp set` with `--force` (with `--oauth`: the same two commands with `--auth oauth` and no header) | `KAGURA_API_KEY=…` in `$OPENCLAW_STATE_DIR/.env` (default `~/.openclaw/.env`), added with an editor |\n\n`--api-key-env VAR` renames `KAGURA_API_KEY` for Codex and OpenClaw\n(an upper-case letter or `_`, then upper-case letters, digits or `_`).\nHermes names its variable itself, so `setup hermes` refuses the option\n(exit 2). The entry goes into Codex's `$CODEX_HOME/config.toml` (default\n`~/.codex/config.toml`), into Hermes's `config.yaml` in `$HERMES_HOME` or\nelse the active Hermes profile's directory (`~/.hermes/profiles/<name>`\nwhen `~/.hermes/active_profile` names one, else `~/.hermes`), or into\nOpenClaw's `$OPENCLAW_CONFIG_PATH` (default `openclaw.json` in the state\ndirectory). The closing notes name the command that checks it:\n`codex mcp get kagura-memory`, `hermes mcp test kagura-memory` or\n`openclaw mcp doctor kagura-memory --probe`. `--mcp-url` defaults to the\nconfigured `mcp_url`, then `https://memory.kagura-ai.com/mcp`. It is read\nas a URL parser reads it, as in Python (python-sdk#279): whitespace and\ncontrol characters around it, and any tab or newline inside it, are\ndropped before the checks, and the entry gets that URL. A plain `http://`\n`--mcp-url` other than localhost, in any spelling, is a usage error\n(exit 2), since the entry sends the key there with every request; so is\nanything but an http(s) URL with a host (`--mcp-url=--help`,\n`memory.kagura-ai.com/mcp`, `https://[::1/mcp`), which would go on the\nharness's command line after `--url`. A configured `mcp_url` gets the same\nchecks, and exits 1 saying to pass `--mcp-url`. `$OPENCLAW_STATE_DIR`,\n`$OPENCLAW_CONFIG_PATH` and `$OPENCLAW_WORKSPACE_DIR` are read as OpenClaw\nreads them: stripped, with a leading `~` expanded.\n\n**The OAuth URL form** (memory-cloud 0.77.0+). From 0.77.0 memory-cloud's\ndynamic client registration accepts the three harnesses' own OAuth clients\non a loopback redirect\n([memory-cloud#1657](https://github.com/kagura-ai/memory-cloud/issues/1657)),\nso `--url-form --oauth --mcp-url https://memory.kagura-ai.com/mcp/w/<workspace-id>`\nwrites a URL entry with no key, no header and no key variable: the harness\nregisters its own client, signs in itself and keeps the token in its own\nstore. Setup never sees the token, never runs a harness `login`, and never\npicks this form on its own. Before it detects, runs or writes anything,\na real run sends one unauthenticated `GET /api/v1/system/info` to the\n`--mcp-url` server and stops (exit 1) unless it reports 0.77.0 or later; a\nversion setup cannot read (no answer, not a 200, unparseable) stops it\ntoo. `--dry-run` sends no request. `--oauth` needs `--url-form` and\n`--mcp-url` and refuses `--api-key-env` (exit 2).\n\n```bash\nnpx kagura-memory setup codex --url-form --oauth --mcp-url https://memory.kagura-ai.com/mcp/w/<workspace-id>\n```\n\n| | The entry | Written with | Sign in | The harness keeps the token in |\n|---|---|---|---|---|\n| Codex | `url` only (Codex's `auth` defaults to OAuth) | `codex mcp add <name> --url <url>`, which saves the entry and then starts Codex's browser sign-in, so setup runs it attached to your terminal (its output on stderr), only with a terminal on stdin and without `-y`; otherwise it prints the table and edits nothing (`--force`: the same add, which signs in again) | `codex mcp login <name>` (`--no-browser` when the browser cannot reach the callback) | the OS keyring (`Codex MCP Credentials`; on Windows, Codex's encrypted secrets store in `~/.codex`), else `~/.codex/.credentials.json` (`$CODEX_HOME`), keyed on the entry's URL |\n| Hermes Agent | `url` + `auth: oauth` | `hermes mcp add <name> --url <url> --auth oauth --connect-timeout 315`, attached, with a terminal and without `-y`: its probe runs the browser sign-in, which its default 30 s bound would cut short. Setup then reads the entry back with `hermes config get mcp_servers.<name> --json`: an entry Hermes kept, saved without `auth: oauth`, or saved disabled (`enabled: false`, after the sign-in did not finish) stops setup (exit 1) with the command that fixes it, and skips the `AGENTS.md` export | `hermes mcp login <name>` (the browser flow), or on memory-cloud 0.78.0+ `hermes mcp login <name> --flow device` (a code entered at the server's `/device` page; no loopback callback, [memory-cloud#1671](https://github.com/kagura-ai/memory-cloud/issues/1671)) | `~/.hermes/mcp-tokens/<name>.json` (`$HERMES_HOME`, or the active Hermes profile's) |\n| OpenClaw | `url`, `transport: \"streamable-http\"`, `auth: \"oauth\"` | `openclaw mcp add <name> --url <url> --transport streamable-http --auth oauth`, which saves an OAuth entry without probing (`--force`: `openclaw mcp set`, even when no entry exists, where Python then uses `mcp add`) | `openclaw mcp login <name>` (`--code <code>` when the browser cannot reach the callback), then `openclaw mcp doctor <name> --probe` | its state database, `~/.openclaw/state/openclaw.sqlite` (`$OPENCLAW_STATE_DIR/state/`) |\n\nmemory-cloud's consent screen shows the client name the harness sends,\nwhich nothing verifies: approve only a sign-in you started. For Codex,\n`-c` or `--guardrails` goes on the URL as `?guardrails=`; Codex keys its\ntoken on the URL, so changing it later means signing in again. With the\nplugin's hooks on, an `--oauth` entry does not get `?guardrails=off` (the\nhooks cannot read it), and setup says so. The `--guardrails` preview runs\non this CLI's own credential when that is on the `--mcp-url` server;\notherwise it names the stored profiles there and runs on the first, or,\nwhen there is none, names the `kagura-memory auth login --server` to run\nfirst. What Codex receives depends on the account it signed in with.\n\nWhat has been checked (python-sdk#284): on 2026-09-25 each harness signed\nin end to end on a `/mcp/w/<workspace-id>` URL against memory-cloud\n0.78.0, through entries the Python CLI's `setup … --url-form --oauth`\nwrote, and reached `tools/list`: Codex 0.157.0, Hermes Agent v2026.9.24\n(the browser sign-in and `hermes mcp login --flow device`) and OpenClaw\n2026.9.6. This bin writes the same entries with the same harness commands.\n\nThis package has no TOML, YAML or JSON5 parser, so it never rewrites those\nfiles. When the harness's CLI is not on `PATH`, the block is printed on\nstderr with the file it belongs in, and stdout stays one JSON document.\nWhen Hermes's `config.yaml` already has an `mcp_servers:` key, only the\n`kagura-memory` entry is printed, to go under it: a second top-level\n`mcp_servers:` would replace the first, and every server in it. An entry\nof the same name, which setup finds by scanning the file, stops the run\n(exit 1, nothing changed) unless you pass `--force`. For Codex that holds\nalways; for Hermes and OpenClaw only with their CLI on `PATH`, since the\nPython CLI finds their entries only through it. Without the CLI, setup\nprints the block to go in place of the old entry and exits 0. A\n`config.toml` that cannot be read stops `setup codex` (exit 1), as in\nPython. A Hermes `config.yaml` that cannot be read, or is not UTF-8,\nstops nothing: setup prints the whole block with Python's note to put\nonly the entry under an `mcp_servers:` key the file may have. An OpenClaw\nfile that cannot be read stops nothing either, since Python never reads\nit, and a note says setup could not look there.\nOn Windows, a CLI installed only as an npm `.cmd` shim counts as not\nfound: Node runs one only through a shell, which would re-parse the\narguments.\n\nWith `--scope user`, the Claude Code entry does not hold the key: it\nsends `Authorization: Bearer ${KAGURA_MCP_API_KEY}`, which Claude Code\nfills in from its own environment each time it connects. The key is never\non the `claude mcp add-json` command line, where any local user could read\nit in the process list, and never in `~/.claude.json`. Set the variable in\nthe environment that starts Claude Code, e.g. `export\nKAGURA_MCP_API_KEY=kagura_xxx` in your shell profile; setup says whether\nthe current shell has it, and `doctor` warns when it is unset. It is a\nseparate variable from `KAGURA_API_KEY`, which the SDK ranks above\n`.kagura.json` and OAuth profiles for every command. The entry is the one\nthe Python CLI writes, so a run of either CLI finds the other's entry up\nto date. `--scope project` still writes the key into `.mcp.json`, so keep\nthat file out of version control (setup adds it to `.gitignore`). A\nuser-scope entry this CLI wrote before 0.11.0 still holds the key in\n`~/.claude.json`: re-run `setup claude --scope user` with\n`KAGURA_MCP_API_KEY` exported to replace it (`doctor` warns about one).\n\n`--guardrails <context-id|off>` (memory-cloud v0.74.0+) sets the URL's\n`guardrails` parameter on `setup claude` and `setup codex`. On `setup\ncodex`, when neither the flag nor the MCP URL (`--mcp-url` or the\nconfigured `mcp_url`) sets one, it defaults to `off` while the Kagura\nplugin's Codex hooks are on for the entry (their `config.json` names its\ntable in `mcp_server`, `kagura-memory` by default), and otherwise to\n`-c`. With a context there, the notes give the `kagura-memory guardrails\ndigest … --target instructions` command, which previews what Codex\nreceives. Hermes and OpenClaw do not read the server's instructions.\nThere `--guardrails off` is refused (exit 2), and a context id is not\nwritten, whether it comes from the flag or from the URL: one warning\nnames what was dropped. A first `?guardrails=off` in the URL, the value\nthe server reads, is kept, alone and with a warning of its own. Names\ncompare decoded, as the server reads them (`guard%72ails` is\n`guardrails`). All of this is as in Python.\n\n**What `tools/list` shows** (the entry's tool view). Since memory-cloud\n0.93.0 a bare MCP URL lists the `core` profile, the memory loop (21 tools\non 0.141.0), and `?profile=full` every tool (72). From memory-cloud\n0.158.0, `?profile=google` and its alias `?profile=microsoft` list `core`\nwithout `describe_tools` and `guide` (`?add_tools=` cannot bring them back),\nwith tool descriptions that name no other tool and instructions that carry\nno guardrail digest. Otherwise `tools/call` is unaffected: a profile is a\nview, not an authorization boundary, so an unlisted tool stays callable and\nevery `KaguraClient` method works on any URL — except that on a `google` /\n`microsoft` URL (without a `?tools=` allowlist, which wins over the\nprofile) `describe_tools` and `guide` are refused at `tools/call` too, so\n`describeTools()` and `guide()` throw `KaguraError` with `code`\n`unknown_tool` there. That view also trims replies: the server drops the\nsentences of its own `help`, `hint` and `message` text that point at another\ntool, and `recall` makes no learning writes. `--tool-profile NAME` (every harness; memory-cloud\nv0.73.0+) sets `profile`; the server knows `core`, `full`, `google` and\n`microsoft` (case-sensitive; the last two need v0.158.0+) and fails\n`tools/list` for any other name, which leaves the harness with no Kagura\ntools. Setup leaves that check to the server and refuses only a blank name\nor one starting with `-` (exit 2), as Python does.\n`--add-tools NAME[,NAME…]` (every harness; v0.129.0+; may\nrepeat) lists more tools on top of the profile: trimmed, de-duplicated,\nat most 100, unknown names ignored by the server. `recall`'s\n`explore_hints` need `explore` listed (v0.135.0 took it out of `core`),\nso add `--add-tools explore` or use `--tool-profile full`. A `?tools=`\nallowlist already on the URL wins over both, which gets a warning.\n`--context CONTEXT_ID` (`setup claude`; v0.114.0+) names the connection's\ndefault context: context-scoped tools called without `context_id` run in\nit, an explicit `context_id` wins, and the default is checked like one,\nso it never widens access; unlike `--context-id`, it must be a UUID. On\ncodex, hermes and openclaw a `?profile=` or `?add_tools=` already in\n`--mcp-url` is kept when the flag is absent, and Codex's `--guardrails`\npreview repeats the entry's view (`--profile` / `--tools` /\n`--add-tools`), so its truncation note names a tool the entry lists. The\nkeys go at the end of the query in Python's order (`guardrails`,\n`profile`, `add_tools`, `context`), replacing any value already there,\nwith commas left literal, as the Python CLI writes them. A `setup claude`\nrun that leaves out a `guardrails`, `profile`, `add_tools` or `context`\nvalue the entry it replaces had, or its `?tools=` allowlist (which goes\nback on `--mcp-url`), says so in a note, reading the old URL as the\nserver does: the first value of each key, trimmed; a blank one, or a\n`?tools=` naming no tool, is left out, as the Python CLI 0.44.0 does.\nServers older than a\nparameter ignore it. `--name`, `--force` and\n`--dry-run` (codex, hermes, openclaw) name the entry, replace an existing\none, and show the command or block without changing anything. A name is\n1-64 letters, digits, `-` or `_`, starting with a letter or digit, as in\nPython: it is a bare argument to `codex` and `openclaw`, which would read\n`--help` as an option. Their `--context-id` must be a context UUID\n(exit 2 otherwise, a padded or empty one included), as in Python without\n`--profile`: this port lists no contexts. `--dry-run` shows one not in\nthe dashed form (dashless, `{braced}`, `urn:uuid:`) as `<UUID of context\nVALUE>`, as Python 0.44.0 does.\n\n**The AGENTS.md export.** `--agents-md [PATH]` on `setup codex`, `setup\nhermes` and `setup openclaw` (memory-cloud v0.74.0+) fetches the\ncontext's tool guardrail export block (`MemoryClient.getGuardrailDigest`)\nand splices it into a file the harness loads every session, between its\n`kagura-memory:guardrails` marker lines: an earlier block is replaced in\nplace, an unchanged set rewrites nothing, and the rest of the file is\nkept, line endings and a symlink included. Without PATH the file is the\nharness's own: `$CODEX_HOME/AGENTS.md` (`AGENTS.override.md` when that\nexists); for Hermes the file Hermes loads from the current directory, found as\nHermes finds it: the nearest `.hermes.md` or `HERMES.md` up to the git\nroot (an empty one ends that search), else this directory's\n`AGENTS.override.md`, `AGENTS.md` or `agents.md` (a new `AGENTS.md` when\nonly a parent directory's loads), else its `CLAUDE.md` or `claude.md`,\nelse a new `AGENTS.md`, so your own file keeps loading. With only Cursor\nrules (`.cursorrules`, `.cursor/rules/*.mdc`) there is no default, since a\nnew `AGENTS.md` would stop Hermes loading them: `--agents-md` then needs a\nPATH (exit 1, nothing run), and a dry run says the export is not offered;\nfor OpenClaw `AGENTS.md` in its workspace,\n`$OPENCLAW_WORKSPACE_DIR`, else `workspace/` in the state directory. A\nleading `~` or `~/` in PATH is expanded to your home directory (a\n`~user` is left as written, as in the Python CLI), and missing\ndirectories are created. The context is `--context-id`, else a\n`--guardrails` context UUID, never the\nURL's `?guardrails=` or `.kagura.json`; without one, `--agents-md` is a\nusage error (exit 2). The block is fetched on the usual credential chain\n(`KAGURA_API_KEY`, the OAuth profile, `.kagura.json`), which setup\nsettles before it runs anything, a dry run included: no credential, or\none for another server than the entry's, stops setup (exit 1) with\nnothing run. The export comes after the entry is applied or printed.\n`wrote` lists the file, and the notes carry Python's lines: the refresh\ncommand (`kagura-memory guardrails digest <ctx> --out <file>`), a warning\npast the 32 KiB Codex reads or the 20,000 characters OpenClaw reads, and\nHermes's prompt-injection note. An empty digest removes an earlier block,\nas `guardrails digest --out` does, and otherwise writes nothing, creating\nno file or directory. A failed export prints the report, then the error\n(exit 1). `--dry-run`\nnames what the export would do to the file (create it, append the block,\nreplace the block, or update a file it cannot read). On Hermes and\nOpenClaw a run without the export ends with Python's \"Re-run with\n--agents-md --context-id <id> …\" hint, except where Hermes has no default\nfile.\n\nClaude Code uses the `kagura-memory` entry from the strongest scope\n(local > project > user). It keys local scope by the git repository root\n(a linked worktree's main working tree), and takes project scope from the\nclosest `.mcp.json` that defines the entry, in the directory it runs in or\nany parent; `setup claude` and `doctor` read them the same way.\n`setup claude` writes nothing when a stronger scope already defines one,\nand prints the `claude mcp remove --scope …` command for it, after a `cd`\ninto the directory that command must run in when that is not the current\none. An entry the new one hides, in a weaker scope or in a parent's\n`.mcp.json`, is noted. Messages name `~/.claude.json` by the file actually\nread: `$CLAUDE_CONFIG_DIR/.claude.json` when that is set.\nWith `--scope user`, an identical user-scope entry is left as it is and a\ndifferent one is replaced: `claude mcp remove` runs before `claude mcp\nadd-json`, and if the add then fails, the old entry is put back. An old\nentry that holds a key (as this CLI wrote before 0.11.0) is never put\nback, since that would pass the key on a command line; when it is not put\nback, or putting it back fails, setup prints the command that re-adds it,\nthe key masked as `<your-api-key>`.\nWithout `claude` on `PATH`, `--scope user` prints the commands to run, and\nwhere to set `KAGURA_MCP_API_KEY`, and writes nothing, unless the\nuser-scope entry is already identical. When `claude plugin list --json`\nshows the Kagura Memory plugin enabled, the notes list the plugin settings\nto enter, and say that the plugin has one guardrail context for every\nproject and authenticates only with a user API key. The printed\n`server_url` is the MCP URL without its query, except that `guardrails=off`\nstays and, from memory-cloud v0.149.0 and the plugin of that release,\n`context=<id>` stays when the MCP URL's first `context=` is the context\nprinted as the plugin's `context_id` (both in the plugin hook's dashed UUID\nform, equal in any case; written lower-case): the plugin's SessionStart hook\nthen tells the model the default context, so it needs no `list_contexts`\ncall. When setup has no plugin context (no `-c`, no `context_id` in the\nproject's `.kagura.json` and no `KAGURA_CONTEXT_ID`), the context is not a\ndashed UUID (a name, or a dashless, braced or `urn:uuid:` spelling, which the\nhook does not accept), or the `context=` names another context, the\n`context=` is dropped. A plugin before v0.149.0 ignores it. `setup claude` writes\n`.kagura.json` and the MCP entry only; it installs none of the Python\nCLI's hooks or `/kagura-recall` and `/kagura-remember` commands. Their\nflags (`--[no-]session-hook`, `--[no-]sync-hook`, `--[no-]commands`) and\n`--no-auto-context` are accepted, so a script written for the Python CLI\nstill runs, and change nothing. `doctor` reports the entry Claude Code\nuses in the current directory, with its scope and file, and warns about\neach entry that one hides. It also warns when a header of that entry\nsends a `${VAR}` that is unset in the current environment, or when that\nentry is in user or local scope and holds the key itself, in\n`~/.claude.json`, saying how to replace it; and for a\n`kagura-mcp` stdio entry (the Python CLI's `--profile` form) checks that\n`kagura-mcp` is on `PATH`. For a `type: \"url\"` entry, which earlier\nreleases wrote and Claude Code skips, it names the fix for the entry's\nscope.\n\n**Not ported.** `kagura ingest` needs the text-extraction pipeline (PDF,\nOffice, EPUB, audio), which does not exist in this package and would cost\nthe zero-dependency promise; use the Python CLI for it. The Claude Code\nextras of `kagura setup claude` are not ported either: its SessionStart\nand PostToolUse hooks and its `/kagura-recall` and `/kagura-remember`\ncommands. `setup claude` here writes `.kagura.json` and the MCP entry\nonly, and takes their flags as inert (see above).\n\nNor are the skills of the Python CLI's Claude Code plugin, such as the\n`memory` skill of 0.42.0 (python-sdk #248), which walks an agent through\n`remember`, `recall`, `reference`, `update-memory` and `forget` and their\n`--details`/`--location` rules: this package ships no Claude Code plugin.\n\nEvery other command of the Python CLI 0.44.0 is here, with the\ndifferences in options and output this section lists: all but\n`bootstrap`, `changes` and `list` as of 0.12.0\n([#57](https://github.com/kagura-ai/kagura-memory-typescript-sdk/issues/57)),\nand those three, which the Python CLI added in 0.44.0, as of 0.15.0\n([#86](https://github.com/kagura-ai/kagura-memory-typescript-sdk/issues/86)).\n\n**Only in this bin.** `secret keygen --reveal`: keygen prints the private\nkey (see key custody below) and refuses to print it to a terminal without\n`--reveal`. `-c` is short for `--context-id` on every `setup` subcommand.\n`auth refresh --no-browser`: a refresh that widens\n`--scope` re-runs the device flow here too, and this skips opening the\nbrowser, as on `auth login`; Python's `auth refresh` takes only\n`--profile` and `--scope`. `-h` is short for `--help` everywhere, where\nthe Python CLI takes only `--help`. `setup codex`, `setup hermes` and\n`setup openclaw` still\naccept `--api-key` and `--project-dir`, which they took before 0.11.0, so\nolder scripts still run. Neither does anything now, and a note says so.\n`--api-key` beside `--profile` is still the usage error (exit 2) it was,\nwith `--url-form` too.\n\n**Five deliberate divergences.**\n\n- `secret` needs the optional `age-encryption` peer (`npm install\n  age-encryption`), exactly as Python's `[secret]` extra works. Key\n  custody differs: Python uses the OS keychain via `keyring`, which has no\n  zero-dependency equivalent in Node, so this CLI reads the age identity\n  from `KAGURA_AGE_IDENTITY` or `KAGURA_AGE_IDENTITY_FILE` and fails\n  closed when neither is set. **A key custodied by the Python CLI is not\n  readable here, and vice versa.**\n- `setup … --profile` (the OAuth path) writes an entry that launches\n  Python's `kagura-mcp` stdio proxy, which this package does not install.\n  Every `setup` subcommand reports the fact, and names the Python CLI's\n  `kagura setup <harness> --profile` command, instead of writing a config\n  that would fail at launch. The `--api-key` path of `setup claude` works\n  here, and so does the URL form of the other three. With `--url-form`,\n  those three ignore `--profile`, with a note: the Python CLI uses it\n  there to check the login, list contexts and fetch the `AGENTS.md`\n  export with that profile alone. This port lists nothing and checks no\n  login, and fetches the export on the usual credential chain, where\n  `KAGURA_PROFILE` picks a profile.\n- `config show` does not reproduce Python's key mask\n  (`key[:8] + \"...\" + key[-4:]`), whose halves overlap below 12 characters\n  and print the whole secret twice.\n- `--progress rich`, and `-v` without `--progress`, print the Python CLI's\n  `→` / `✓` / `✗` lines as plain text, the text Rich writes when stderr is\n  not a terminal: no colours even on a terminal, no wrapping at its width,\n  and no Rich markup, so `report[bold].pdf` prints as written (Rich drops\n  the `[bold]`, and a `[/…]` in an error message makes it fail). `-vv` and\n  `-vvv` show one line per detail or debug event, where Rich draws a panel;\n  neither command emits one. `--progress json` is the same NDJSON.\n- With no `.kagura.json`, the MCP-backed commands (`recall`, `guardrails\n  load`, `measure`, …) and `doctor` reach an OAuth profile's own server.\n  The Python CLI takes the default URL its config loader fills in for the\n  server, and sends the profile's token there. A `KAGURA_API_KEY` goes to\n  `KAGURA_MCP_URL`, else the default, in both.\n\n**`setup codex`, `setup hermes` and `setup openclaw`** also differ from\nthe Python CLI in these ways, each on purpose:\n\n- Every entry here is the URL form, so `--url-form` is accepted and\n  changes nothing by itself (`--oauth` needs it, as in Python), and\n  `--mcp-url` falls back to the configured URL where Python requires it\n  (not with `--oauth`). Only that fallback needs the configuration: without\n  `--mcp-url`, a `.kagura.json` that cannot be loaded stops setup (exit 1);\n  with it, setup goes on with a note, as Python never reads that file.\n- `-c` must be a context UUID with `--url-form --profile` too (exit 2),\n  where Python looks a name up through the profile.\n- `setup hermes` never runs `hermes mcp add` for the API-key form, which\n  prompts for the key; it prints the block, as Python does under `-y`.\n  With `--oauth` it runs it attached, as Python does.\n- So nothing is read back with `hermes config get` for that form either:\n  the Python CLI 0.41.1's stops after the add (an entry Hermes kept in\n  place of the one asked for, an entry saved disabled, which points to\n  `hermes mcp test`) and its warning for a URL entry saved without an\n  `Authorization` header cannot arise for the printed block.\n- An existing entry is found by scanning the file, and not described by\n  kind. A readable but malformed `config.toml` does not stop\n  `setup codex`.\n- On Codex, a `?guardrails=` already in the MCP URL (`--mcp-url` or the\n  configured `mcp_url`) beats the hooks and `-c` defaults.\n- Nothing prompts here (setup only hands the terminal to an `--oauth`\n  `codex mcp add` or `hermes mcp add`), so `--agents-md` always needs a\n  context, and the export is never offered: a dry run of Hermes or\n  OpenClaw says \"not offered: this port never prompts\" (or \"not offered\n  with -y\").\n- When the export credential is for another server, the message\n  says what moves it here, where Python's points at `--profile`: for\n  `KAGURA_API_KEY`, which outranks every profile, `KAGURA_MCP_URL` or\n  unsetting the key; otherwise `KAGURA_PROFILE`.\n- File errors of the export are Node's words, not Python's `[Errno …]`.\n- Output is one JSON document on stdout, with Python's sentences in\n  `notes` and the block on stderr. A non-ASCII URL is written as UTF-8\n  where Python writes `\\uXXXX` escapes; both are valid.\n- `--oauth`: a harness CLI run attached writes its output to stderr, so\n  stdout stays one JSON document. The server-check refusal ends \"Use\n  --url-form with an API key instead (no --oauth).\", where Python also\n  offers its stdio entry, which this bin cannot write. The Codex hooks\n  warning appears only when the plugin's hooks are on for the entry (this\n  bin does not read an existing entry's bearer), and setup does not ask\n  \"Write the --oauth entry anyway?\": it warns and goes on, as Python does\n  under `-y`. The `--guardrails` preview never names `--profile`, which is\n  inert here, and also names a `.kagura.json` that is JSON but not an\n  object, which every command of this bin refuses (Python's loader takes\n  one). Sentences Python wraps at 78 columns are one line here.\n\nSome small divergences run the other way: this CLI refuses what click\nwould accept.\n\n- `--profile=` and `--scope=` reject an explicitly empty value. An empty\n  profile name would create a nameless profile and an empty scope would go\n  to the server verbatim. Other options take `--flag=` as Python does,\n  apart from `--external-id=` in the `update-memory` case below.\n- `context update --lock --unlock` is a usage error (exit 2, \"mutually\n  exclusive; pick one\"), where click takes whichever flag comes last.\n- `--rerank --no-rerank` is the same usage error, on `recall` and on\n  `context search-config`, and so are `--content-chunks\n  --no-content-chunks` and `--llm-selector --no-llm-selector`.\n- `auth logout --all --profile NAME` is the same usage error, where click\n  ignores `--profile` and logs out every profile.\n- `update-memory --dismiss-supersede-candidate`, and the other flags\n  that need `--memory-id` (`--accept-supersede-candidate`,\n  `--merge-details`, `--details-patch`, `--add-tag`, `--remove-tag`,\n  `--expected-version`), exit 1 beside any `--external-id`, before\n  anything is sent. The Python CLI refuses only a non-empty one, and sends\n  an empty `--external-id=` to the server beside `-m`, which refuses both\n  ids; this CLI refuses that too, as `updateMemory` does.\n- `-k=5` (a short option, `=`, a value) reads `5`, where click reads `=5`\n  as the value and then refuses it.\n- An option that takes a value does not take a following argument that\n  begins with a dash and is not a number (a lone `-` is still a value):\n  `--name-contains -auth` is a missing value here (`Error: Option\n  '--name-contains' requires an argument.`), where click takes `-auth`.\n  Write `--name-contains=-auth`. A number is anything `float()` reads,\n  so `-١` is one. `auth login --invite` is the exception, since an invite\n  token may begin with a dash.\n- An id the command puts in a URL path, `FILE_ID` (`files delete`, `files\n  download-url`), the resource id (`-r/--resource-id` of `resource stats`,\n  `indexer-status`, `schema`, `ingest`, `ingest-batch` and `import`, and\n  `resource events RESOURCE_ID`) and `USER_ID` / `--user`, is refused when\n  it is `.`, `..` or empty (exit 2, `Invalid value for 'FILE_ID': '..' is\n  not a valid file id.`), before anything is read or sent. The Python CLI\n  sends it, and URL resolution then drops or climbs the segment: `files\n  download-url ..` asks for `/api/v1/download-url`. (From 0.41.1 it\n  refuses a `USER_ID` as this bin does.) Any other id is sent\n  percent-encoded as one segment, so a `/`, `?` or `#` in it stays in the\n  id rather than reaching another route or replacing the query\n  (`files delete 'x?workspace_id=…'`).\n\n**Differences from the Python CLI 0.42.0–0.44.0 in what those releases\nadded**, each on purpose or out of reach:\n\n- `update-memory --details` and `remember --details` refuse a `NaN` or\n  an infinity in the object (`Out of range float values are not JSON\n  compliant: nan`, exit 1, Python's line) before the context is looked\n  up; the Python CLI reads the object as JSON and fails only when it\n  sends it, so without a context it reports that first. (`--details-patch`\n  refuses it where Python does, never sending a `null` that would delete\n  the key.) Text that is not JSON at all gets Python's `json` module's\n  message, as `list --filters` does.\n- `update-memory --merge-details` prints the size of a bounded\n  `reference` reply when `details_total_chars` is a whole number sent as a\n  float (`24000.0`), which Python leaves out: JSON.parse cannot tell it\n  from `24000`.\n- A message of `update-memory` (0.44.0) that names a command says\n  `kagura-memory reference`, where the Python CLI's says `kagura\n  reference`.\n- `getServerInfo`, `checkServerVersion`, `getEmbeddingStatus`,\n  `getMemoryStats`, `findDuplicates` and `listEmbeddingModels` check the\n  body against the Python model and return it as the server sent it:\n  keys the model does not have stay, and a lax value such as `\"3\"` for an\n  int is not converted, where Python returns the model.\n- `getMemoryStats` accepts rows without `use_count`, which memory-cloud\n  v0.34.0 and later never send. The Python SDK's model (0.44.0 too) still\n  requires it, so it refuses every non-empty page\n  (`memories.0.use_count: Field required`).\n- `listMemories` returns its body unchecked. The Python SDK has read it\n  through `MemoryListResponse` since 0.40.0.\n- The kept Hermes OAuth entry of 0.41.3 (python-sdk #287) cannot arise\n  here. In Python it arises when the API-key form runs `hermes mcp add`,\n  its overwrite prompt is declined, and an `auth: oauth` entry at the\n  same URL stays in place. An `--oauth` request already matches that\n  entry, so it never stops there. This port's API-key form never runs\n  `hermes mcp add`: it prints the entry to put in place of the existing\n  one, as Python does under `-y`.\n- `sleep rollback` prints one line the Python CLI 0.44.0 does not:\n  `related_indexes_reverted` in the summary, after `points_requeued` (0\n  when the server sends no key). It is the counter memory-cloud v0.153.0\n  added for its opt-in Sleep related index, printed here as the Python CLI\n  prints it from 0.46.0; the rest of the output is 0.44.0's.\n- `files list` and `files upload` print a file's `scan_status`, after\n  `context_id`, when the server sends it (memory-cloud v0.160.0+), which the\n  Python CLI 0.44.0 drops; a reply without it prints as 0.44.0 prints it,\n  with no such key. `files download-url` on a file the malware scan holds\n  or blocked prints one line more under the error: how long to wait and to\n  run the command again, or that the file cannot be downloaded (the\n  wording of the Python SDK's\n  [#337](https://github.com/kagura-ai/kagura-memory-python-sdk/issues/337)).\n- `bootstrap` prints the `notices` block (memory-cloud v0.161.0+) right\n  after `status` when the server sends it, which the Python CLI 0.44.0\n  drops (memory-cloud has then already counted the notices as delivered); a\n  reply without it prints as 0.44.0 prints it, with no such key, and an\n  unreadable block prints `null`.\n\n**`guardrails load` and `guardrails digest`** (memory-cloud v0.74.0+)\ntake the context as an optional argument, else `context_id` from\n`.kagura.json`. `load` calls the MCP `load_guardrails` tool and prints the\nset as the Python CLI prints it: the model's sixteen keys in its order\n(`context_summary_omitted` last, `false` when the tool sends none),\nwithout the `context_display_name`, `context_is_private` and\n`context_is_locked` the tool adds, each item's eleven keys, and `null` for\na `tool_trigger` the Python SDK cannot read (`match: null` is filled in\nwhen a trigger has none). A set missing a required field, a truncation\nflag above all, is refused with the Python SDK's `KaguraResponseError`\ntext (exit 1) rather than printed as complete. `--cap` takes 1-1000.\n`digest` reads the REST digest route with the credential chain (an API\nkey or an OAuth profile) and prints the text: the `AGENTS.md` export\nblock, or with `--target instructions` the MCP server instructions this\ncredential would receive, where `--profile`, `--tools` and `--add-tools`\n(memory-cloud v0.129.0+; comma-separated, may repeat) repeat an MCP URL's\n`?profile=`, `?tools=` and `?add_tools=`; without them, memory-cloud 0.93.0+\npreviews the `core` view a bare URL lists. `--profile` takes `full`, `core`,\n`google` or `microsoft`; for the last two (v0.158.0+) the preview is that\nview's base text, with no digest. A context with no tool guardrails prints\nnothing, and says so on stderr. With `--out FILE`, it splices the block\ninto FILE between its marker lines: an earlier block is replaced in place,\nan unchanged set rewrites nothing, and an empty set removes an earlier\nblock. It then prints one JSON line, `{\"path\": …, \"status\":\n\"written\"|\"unchanged\"|\"removed\", \"tool_triggered_version\": …}`, with the\npath as Python's `pathlib` writes it (`./x/../X.md` reads `x/../X.md`).\nAs in Python, `--out` does not expand `~`, never creates a missing parent\ndirectory, and keeps a symlink, the file's mode and its line endings.\nBeside JSON's `1` for Python's `1.0` (an `importance`), two differences\nremain: an item's `created_at` / `updated_at` is printed as\nthe server sent it, where Python reformats it and refuses one it cannot\nparse; and a failed read or write of the `--out` file reads in Node's\nwords (`ENOENT: no such file or directory, open '…'`) rather than\nPython's (`[Errno 2] No such file or directory: '…'`).\n\n**`sleep history` and `sleep report`** print what the Python CLI prints:\n`{\"reports\": […]}` and the report, each read through the Python SDK's\nmodel (`SleepReport`, `SleepReportDetail`) — its keys in its order, its\ndefault for a key an older server leaves out (`source: null`,\n`actions_has_more: false`, an action's `details_omitted: false`), a\ntimestamp in pydantic's form — and a report the model refuses exits 1\nwith its `KaguraResponseError` text. `history --source\nsleep|analysis|all` (memory-cloud v0.138.0+) lists Sleep runs (the\nserver's default), the rows Memory Analysis writes for its LLM spend, or\nboth. `report --actions-limit N` (1-200; the server's default is 50) and\n`--actions-cursor C` (memory-cloud v0.82.0+) page the audit log: while\n`actions_has_more` is true, pass `actions_next_cursor` as\n`--actions-cursor`. `report`'s `CONTEXT_ID` is kept for compatibility\nand not sent: the server finds the run by `REPORT_ID`.\n\n```bash\nnpx kagura-memory sleep history <context-id> --limit 5 --source all\nnpx kagura-memory sleep report <context-id> <report-id> --actions-limit 100 --actions-cursor <cursor>\n```\n\n**`doctor`** checks the credentials, the `kagura-memory` entry Claude Code\nuses, the optional `age-encryption` peer and the age identity, then the\nserver, whose checks are the Python CLI's: `Server reachable`, then\n`Version: <version>` by the same comparison `checkServerVersion` makes\n([#66](https://github.com/kagura-ai/kagura-memory-typescript-sdk/issues/66)).\nThe version passes at or above `MIN_SERVER_VERSION` (0.75.0), fails below\nit (`Version: 0.74.0 is below minimum 0.75.0`, a pre-release of 0.75.0\nincluded) and is `info` when it cannot be compared (`main-abc123`, `0.78`).\nAn unreachable server fails with `Server unreachable: …`, and an OAuth\nprofile the REST route refuses is `info`, in Python's words. The\n`/system/info` body is read through Python's `ServerInfo` model, and one\nit refuses (no `name`, a `version` that is no string) fails as the Python\nCLI 0.42.0 fails it: `Server answered, but the SDK could not read\n/api/v1/system/info: KaguraClient.get_server_info: unexpected server\nresponse for ServerInfo (…). The server may be newer than this SDK;\n","readmeFilename":"README.md"}