{"_id":"@ajinkyakulkarni/cost-calc-mcp","name":"@ajinkyakulkarni/cost-calc-mcp","dist-tags":{"latest":"1.0.0"},"versions":{"1.0.0":{"name":"@ajinkyakulkarni/cost-calc-mcp","version":"1.0.0","description":"MCP server for the AI Cost Calculator — numbers byte-identical to calc.ajinkya.ai","type":"module","bin":{"cost-calc-mcp":"bin/cost-calc-mcp.js"},"engines":{"node":">=18"},"scripts":{"prepare":"node ../scripts/copy-engine.mjs","build":"node ../scripts/copy-engine.mjs","test":"node ../scripts/test-packaged.mjs"},"dependencies":{"@modelcontextprotocol/sdk":"^1.29.0","zod":"^4.0.0"},"keywords":["mcp","llm","ai","cost-calculator","agent","anthropic","openai"],"author":{"name":"Ajinkya Kulkarni"},"license":"MIT","repository":{"type":"git","url":"git+https://github.com/ajinkyakulkarni/ai-cost-calculator-studio.git"},"homepage":"https://calc.ajinkya.ai","_id":"@ajinkyakulkarni/cost-calc-mcp@1.0.0","gitHead":"8f3915c083a107e9087a5312c09be94a087d305b","bugs":{"url":"https://github.com/ajinkyakulkarni/ai-cost-calculator-studio/issues"},"_nodeVersion":"22.14.0","_npmVersion":"10.9.2","dist":{"integrity":"sha512-vJBZHzR9qGr0GKFkXBxE6b1MP8jV/D4hfLH5oh4JvyQ1o6FuG4k815RRCimoPdhPXqUf7T6QQXij+A6SJPh6zQ==","shasum":"fc9827219a721e34609b26d4783cf72ec60071f4","tarball":"https://registry.npmjs.org/@ajinkyakulkarni/cost-calc-mcp/-/cost-calc-mcp-1.0.0.tgz","fileCount":37,"unpackedSize":363056,"signatures":[{"keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U","sig":"MEUCIBCPilwKppWG2FjCJJ28bxSKHlU2tEC/wXa2ny0aSOM2AiEAlYc8bKfwKnlXSDnGQOrCkBFXDRknAtPVHkrcy1Zc3a0="}]},"_npmUser":{"name":"ajinkyakulkarni","email":"npmjs@ajinkyakulkarni.com"},"directories":{},"maintainers":[{"name":"ajinkyakulkarni","email":"npmjs@ajinkyakulkarni.com"}],"_npmOperationalInternal":{"host":"s3://npm-registry-packages-npm-production","tmp":"tmp/cost-calc-mcp_1.0.0_1782862537753_0.5149075665861809"},"_hasShrinkwrap":false}},"time":{"created":"2026-06-30T23:35:37.543Z","1.0.0":"2026-06-30T23:35:37.933Z","modified":"2026-06-30T23:35:38.121Z"},"maintainers":[{"name":"ajinkyakulkarni","email":"npmjs@ajinkyakulkarni.com"}],"description":"MCP server for the AI Cost Calculator — numbers byte-identical to calc.ajinkya.ai","homepage":"https://calc.ajinkya.ai","keywords":["mcp","llm","ai","cost-calculator","agent","anthropic","openai"],"repository":{"type":"git","url":"git+https://github.com/ajinkyakulkarni/ai-cost-calculator-studio.git"},"author":{"name":"Ajinkya Kulkarni"},"bugs":{"url":"https://github.com/ajinkyakulkarni/ai-cost-calculator-studio/issues"},"license":"MIT","readme":"# @ajinkyakulkarni/cost-calc-mcp\n\nAn **MCP server** that costs AI-agent deployments — wired to the same engine that powers **[calc.ajinkya.ai](https://calc.ajinkya.ai)**, so the numbers it returns are byte-identical to the visual calculator.\n\nThe LLM (your Claude Code / Cursor / Claude Desktop) runs the conversation; it **never does the arithmetic**. Every dollar figure comes from `compute_cost`, which is *hard-gated* — it refuses to return a number until the cost-driving inputs (volume, model, hosting, cache rate, per-query token profile) are actually provided. Describe your deployment in plain language; the agent proposes sensible defaults, confirms what it must, computes, and hands back a `calc.ajinkya.ai/#w=…` link to open the full visual model.\n\n## Install\n\n**Claude Code:**\n```bash\nclaude mcp add cost-calc -- npx -y @ajinkyakulkarni/cost-calc-mcp\n```\n\n**Cursor / Claude Desktop** — add to your `mcpServers` config:\n```json\n{\n  \"mcpServers\": {\n    \"cost-calc\": {\n      \"command\": \"npx\",\n      \"args\": [\"-y\", \"@ajinkyakulkarni/cost-calc-mcp\"]\n    }\n  }\n}\n```\n\n### Prefer zero install? Use the hosted endpoint\nNo npm needed — point your client at the hosted Worker:\n```bash\nclaude mcp add --transport http cost-calc https://calc.ajinkya.ai/mcp\n```\n(Cursor / Desktop: an `mcpServers` entry with `\"url\": \"https://calc.ajinkya.ai/mcp\"`.)\n\n## Tools\n`list_presets` · `load_preset` · `get_schema` · `validate_workload` · `compute_cost` · `make_share_link`\n\nPlus a `cost_interview` prompt for a guided session — or just say *\"help me cost an AI agent.\"*\n\n## Why trust the numbers\n- **Same engine as the website.** This package bundles the canonical engine modules verbatim (no fork); a parity check asserts byte-identical output across all 18 presets.\n- **No hallucinated math.** The model only fills in a structured `workload` and calls the engine.\n- **Auditable.** Every result carries the assumptions it made and the engine's derivation trace.\n\n## Requirements\nNode ≥ 18. Runs locally over stdio; no API key (the engine is pure math — your client provides the model).\n\n## Links\n- Calculator: https://calc.ajinkya.ai\n- Source: https://github.com/ajinkyakulkarni/ai-cost-calculator-studio\n\nMIT\n","readmeFilename":"README.md","_rev":"1-83a47284107361a7441d6c757c359295"}