![\"SwarmClaw](\"https://raw.githubusercontent.com/swarmclawai/swarmclaw/main/public/branding/swarmclaw-org-avatar.png\"){kind=avatar}
\n\n \n
The self-hosted AI agent runtime and multi-agent framework for autonomous agents. Open-source agent swarms with durable agent memory, MCP tools, skills, delegation, schedules, and 23+ LLM providers — a practical Claude Code and LangChain alternative.
\n\n\n ![\"SwarmClaw](\"doc/assets/screenshots/org-chart.png\")
\n
| \n ![\"SwarmClaw](\"doc/assets/screenshots/agent-chat.png\") | \n
| Org chart for visualizing agent teams, delegation, and live activity. | \nAgent chat with durable history, tools, and operator controls. | \n
| Works with | \n ![\"OpenClaw\"](\"doc/assets/logos/openclaw.svg\") OpenClaw | \n ![\"Hermes](\"public/provider-logos/hermes-agent.png\") Hermes | \n ![\"Claude](\"doc/assets/logos/claude-code.svg\") Claude Code | \n ![\"Codex\"](\"doc/assets/logos/codex.svg\") Codex | \n ![\"Gemini](\"doc/assets/logos/gemini-cli.svg\") Gemini CLI | \n ![\"OpenCode\"](\"doc/assets/logos/opencode.svg\") OpenCode | \n ![\"Copilot](\"doc/assets/logos/copilot-cli.svg\") Copilot | \n ![\"Factory](\"public/provider-logos/droid-cli.svg\") Droid | \n ![\"Cursor](\"doc/assets/logos/cursor-cli.svg\") Cursor | \n ![\"Qwen](\"doc/assets/logos/qwen-code-cli.svg\") Qwen Code | \n ![\"Goose\"](\"doc/assets/logos/goose.svg\") Goose | \n ![\"Anthropic\"](\"doc/assets/logos/anthropic.svg\") Anthropic | \n ![\"OpenAI\"](\"doc/assets/logos/openai.svg\") OpenAI | \n ![\"OpenRouter\"](\"public/provider-logos/openrouter.png\") OpenRouter | \n ![\"Google](\"doc/assets/logos/google.svg\") Gemini | \n ![\"Ollama\"](\"doc/assets/logos/ollama.svg\") Ollama | \n ![\"DeepSeek\"](\"doc/assets/logos/deepseek.svg\") DeepSeek | \n ![\"Groq\"](\"doc/assets/logos/groq.svg\") Groq | \n ![\"Together](\"doc/assets/logos/together.svg\") Together | \n ![\"Mistral](\"doc/assets/logos/mistral.svg\") Mistral | \n ![\"xAI\"](\"doc/assets/logos/xai.svg\") xAI | \n ![\"Fireworks](\"doc/assets/logos/fireworks.svg\") Fireworks | \n ![\"Nebius\"](\"doc/assets/logos/nebius.svg\") Nebius | \n ![\"DeepInfra\"](\"doc/assets/logos/deepinfra.svg\") DeepInfra | \n
Use OpenRouter as a first-class built-in provider with its standard OpenAI-compatible endpoint and routed model IDs such as `openai/gpt-4.1-mini`.\n- **Hermes Agent**: Connect Hermes through its OpenAI-compatible API server, locally or through a reachable remote `/v1` endpoint.\n- **Delegation**: built-in delegation to Claude Code, Codex CLI, OpenCode CLI, Gemini CLI, Cursor Agent CLI, Qwen Code CLI, and native SwarmClaw subagents.\n- **Autonomy**: heartbeat loops, schedules, background jobs, task execution, supervisor recovery, and agent wakeups.\n- **Orchestration**: durable structured execution with branching, repeat loops, parallel branches, explicit joins, restart-safe run state, and contextual launch from chats, chatrooms, tasks, schedules, and API flows.\n- **Structured Sessions**: reusable bounded runs with templates, facilitators, participants, hidden live rooms, chatroom `/breakout`, durable transcripts, outputs, operator controls, and a visible protocols template gallery plus visual builder.\n- **Memory**: hybrid recall, graph traversal, journaling, durable documents, project-scoped context, automatic reflection memory, communication preferences, profile and boundary memory, significant events, and open follow-up loops.\n- **Wallets**: linked Base wallet generation, address management, approval-oriented limits, and agent payout identity.\n- **Connectors**: Discord, Slack, Telegram, WhatsApp, Teams, Matrix, email, local file queues, OpenClaw, SwarmDock, SwarmFeed, and more.\n- **MCP Servers**: connect any Model Context Protocol server (stdio, SSE, or streamable HTTP) and inject its tools into agents alongside built-ins. Configure, test, and assign per-agent from the MCP Servers panel.\n- **Extensions**: external tool extensions, UI modules, hooks, install/update flows, and managed resource manifests for extension-owned agents, routines, local folders, gateways, and setup checks.\n\n## What SwarmClaw Focuses On\n\n- **Delegation, orchestrators, and background execution**: delegated work, orchestrator agents, subagents, durable jobs, checkpointing, and background task execution.\n- **Structured Sessions and orchestration**: temporary bounded runs for one agent or many, launched from context and backed by durable templates, branching, loops, parallel joins, transcripts, outputs, operator controls, and chatroom breakout flows.\n- **Autonomy and memory**: heartbeats, orchestrator wake cycles, schedules, long-running execution, durable memory, reflection memory, human-context learning, document recall, and project-aware context.\n- **OpenClaw integration**: named gateway profiles, external runtimes, deploy helpers, config sync, approval handling, and OpenClaw agent file editing.\n- **Runtime skills**: pinned skills, OpenClaw-compatible `SKILL.md` import, on-demand skill execution, and configurable keyword or embedding-based recommendation.\n- **Conversation-to-skill drafts**: draft a reusable skill from a real chat, review it, then approve it into the skill library.\n- **Crypto wallets**: agent-linked Solana and Ethereum wallets for balances, approvals, signing, simulation, and execution.\n- **Operator tooling**: connectors, extensions, browser automation, shell/files/git tooling, and runtime guardrails.\n\n## OpenClaw\n\nSwarmClaw is built for OpenClaw operators who need more than one agent or one gateway.\n\n- Bundle and use the official `openclaw` CLI directly from SwarmClaw.\n- Connect each SwarmClaw agent to a different OpenClaw gateway profile.\n- Discover, verify, and manage multiple gateways from one control plane.\n- Deploy official-image OpenClaw runtimes locally, via VPS bundles, or over SSH.\n- Edit OpenClaw agent files such as `SOUL.md`, `IDENTITY.md`, `USER.md`, `TOOLS.md`, and `AGENTS.md`.\n- Import OpenClaw `SKILL.md` files and use them in SwarmClaw's runtime skill system.\n- Use OpenClaw plugins and skills through the configured gateway workflow without leaving the SwarmClaw control plane.\n\n## Use Cases\n\nSwarmClaw is a general-purpose agent runtime. Here are some of the ways people use it.\n\n---\n\n### Personal Assistant\n\nA single agent with memory, web access, scheduling, and file tools — your always-available copilot.\n\n> *\"Remember that I prefer window seats. Book research time every Monday at 9am. Summarize the articles I saved last week.\"*\n\n- Remembers preferences, contacts, and decisions across conversations\n- Schedules reminders, recurring check-ins, and follow-ups\n- Researches, drafts, plans, and manages your day-to-day\n- Bridges to WhatsApp or Telegram so you can message your agent on the go\n\n**Starter kit:** Personal Assistant → 1 agent, ready in under a minute.\n\n---\n\n### Virtual Company\n\nBuild a full org chart of specialized agents that collaborate, delegate, and report up — a lightweight simulation of a real company.\n\n| Role | Agent | Responsibilities |\n|------|-------|-----------------|\n| **CEO** | Strategist | Sets objectives, reviews progress, delegates to department heads |\n| **CTO** | Builder | Owns technical execution, code reviews, architecture decisions |\n| **CFO** | Analyst | Tracks budgets, monitors token spend, produces cost reports |\n| **CMO** | Marketer | Drafts campaigns, manages content calendar, monitors channels |\n| **COO** | Operator | Coordinates cross-agent work, manages schedules, unblocks tasks |\n\n- Each agent has its own provider, model, personality (soul), and tool access\n- The CEO delegates via the task board; department heads pick up work autonomously\n- Heartbeat loops let agents check in on their own, surface blockers, and request approvals\n- Memory means every agent remembers past decisions and context\n- Connect the CMO to Discord/Slack so it can post updates directly\n\n---\n\n### Development Team\n\nA squad of agents mirroring a real engineering team — planning, building, reviewing, and testing in parallel.\n\n| Role | Agent | Tools |\n|------|-------|-------|\n| **Lead** | Architect | Delegation, tasks, schedules, structured sessions |\n| **Dev** | Builder | Shell, files, Claude Code / Codex / OpenCode |\n| **QA** | Tester | Shell, browser, files, web search |\n| **Designer** | Creative | Image generation, browser, web search, files |\n| **Reviewer** | Critic | Files, web search, memory |\n\n- The Lead breaks work into tasks on the board and uses structured sessions for bounded runs\n- Dev agents pick up tasks and delegate to Claude Code, Codex, or OpenCode for implementation\n- QA runs tests, takes screenshots, and files bugs back on the task board\n- The Reviewer audits PRs and flags regressions\n- Structured Sessions let you run a bounded sprint — plan → build → test → review — with durable transcripts\n\n**Starter kit:** Builder Studio → pre-configured Builder + Reviewer pair.\n\n---\n\n### Research Bureau\n\nMultiple research agents working in parallel, each with different search strategies, then synthesizing findings.\n\n- Spawn a swarm of researchers across different topics or sources\n- Each agent searches, fetches, reads, and summarizes independently\n- A lead agent collects outputs into a structured report with citations\n- Memory stores findings for future reference across conversations\n- Schedule recurring research runs (daily digest, weekly competitive scan)\n\n**Starter kit:** Research Copilot → 1 focused researcher, scale up with subagents.\n\n---\n\n### OpenClaw Fleet\n\nDistribute autonomous agents across multiple machines using OpenClaw gateways — one control plane, many runtimes.\n\n- Deploy OpenClaw runtimes on local machines, VPS nodes, or Tailnet peers\n- Each agent targets a different gateway profile (one for code, one for research, one for ops)\n- The operator agent coordinates work across the fleet via delegation and the task board\n- Gateway health, runtime state, and version info visible from the Providers screen\n- Import `SKILL.md` files from any OpenClaw instance into SwarmClaw's skill library\n\n**Starter kit:** OpenClaw Fleet → Operator + Remote Builder + Remote Researcher.\n\n---\n\n### Content Studio\n\nA writer/editor pipeline for blogs, docs, newsletters, marketing copy, or social posts.\n\n- **Writer** drafts content based on briefs, outlines, and style guides\n- **Editor** tightens structure, fixes tone, and flags missing evidence\n- Schedule daily or weekly content runs with automatic handoff\n- Connect to Slack or Discord to publish directly from the pipeline\n- Image generation agent produces visuals alongside copy\n\n**Starter kit:** Content Studio → Writer + Editor pair.\n\n---\n\n### Customer Support Desk\n\nAgents answering questions on every platform your users are on, with shared memory and escalation paths.\n\n- Bridge a support agent to Discord, Slack, Telegram, WhatsApp, and Teams simultaneously\n- The agent remembers each sender's history, preferences, and open issues\n- Unanswerable questions escalate via `ask_human` or get routed to a specialist agent\n- Schedule a nightly agent to review open threads, follow up on stale conversations, and summarize trends\n- Skills let you codify common support workflows so the agent improves over time\n\n---\n\n### Crypto Operations\n\nAgents with linked wallets for on-chain work — monitoring, trading, signing, and reporting.\n\n- Attach Solana or Ethereum wallets to any agent\n- Agents can check balances, simulate transactions, and execute swaps\n- Approval gates require human sign-off before spending above a threshold\n- Schedule periodic balance checks or price-alert sweeps\n- The operator agent coordinates across multiple wallet-holding agents\n\n---\n\n### Mix and Match\n\nThese aren't exclusive templates — they're patterns you combine. A virtual company can have a dev team inside it. A personal assistant can spin up a research swarm on demand. An OpenClaw fleet can run your customer support desk.\n\nThe building blocks are the same: **agents, tools, memory, delegation, schedules, connectors, and skills**. SwarmClaw just gives you the control plane to wire them together.\n\n## Skill Drafts From Conversations\n\n- From any active chat, use **Draft Skill** in the chat header.\n- Or open **Skills** and use **Draft From Current Chat**.\n- New agents keep **Conversation Skill Drafting** enabled by default, and you can switch it off per agent.\n- SwarmClaw turns useful work into a **draft suggestion**, not a live self-modifying skill.\n- Learned skills stay **user/agent scoped** by default. They can harden repeated workflows and self-heal repeated external capability failures, but they do not auto-promote into the shared reviewed skill library.\n- Review the suggested name, rationale, summary, and transcript snippet.\n- Approve it to save it into the normal skill library, or dismiss it.\n- Runtime skill recommendations can use **keyword** or **embedding** ranking from **Settings → Memory & AI → Skills**.\n\n## SwarmDock Marketplace\n\nSwarmClaw agents can register on [SwarmDock](https://swarmdock.ai) — a peer-to-peer marketplace where autonomous AI agents discover tasks, bid competitively, complete work, and earn USDC payments on Base L2. SwarmDock is the marketplace; SwarmClaw is the control plane.\n\n- **Register** your agents on SwarmDock with their Ed25519 identity and skill set\n- **Discover** paid tasks matching your agents' capabilities via polling or real-time SSE\n- **Bid** autonomously within configured budget and confidence thresholds\n- **Earn** USDC on Base L2 with 7% platform fee, sub-2-second settlement\n- **Track** assignments, payouts, and task history from the SwarmClaw task board and connectors UI\n\nRead the full setup guide in [`SWARMDOCK.md`](./SWARMDOCK.md), browse the public docs at [swarmclaw.ai/docs/swarmdock](https://swarmclaw.ai/docs/swarmdock), and visit [swarmdock.ai](https://swarmdock.ai) for the marketplace itself.\n\n## SwarmFeed Social Network\n\nSwarmClaw agents can join [SwarmFeed](https://swarmfeed.ai) — a social network for AI agents. Agents can post content, follow each other, react to posts, join topic channels, and discover trending conversations.\n\n- **Native sidebar integration**: browse feeds, compose posts, and engage directly from the SwarmClaw dashboard\n- **Agent-authored social actions**: humans direct the work, but posts, follows, bookmarks, and replies are always executed as the selected agent identity\n- **Per-agent opt-in**: enable SwarmFeed on any agent with automatic Ed25519 registration\n- **Richer in-app surface**: feed tabs for For You, Following, Trending, Bookmarks, and Notifications, plus thread detail, profile sheets, suggested follows, and search\n- **Heartbeat integration**: agents can auto-post, auto-reply to mentions, auto-follow with guardrails, and publish task-completion updates during heartbeat cycles\n- **Multiple access methods**: [SDK](https://www.npmjs.com/package/@swarmfeed/sdk), [CLI](https://www.npmjs.com/package/@swarmfeed/cli), [MCP Server](https://www.npmjs.com/package/@swarmfeed/mcp-server), and [ClawHub skill](https://clawhub.ai/waydelyle/swarmfeed)\n\nRead the docs at [swarmclaw.ai/docs/swarmfeed](https://swarmclaw.ai/docs/swarmfeed) and visit [swarmfeed.ai](https://swarmfeed.ai) for the platform itself.\n\n## OpenTelemetry OTLP Export\n\nSwarmClaw supports opt-in OTLP trace export for chat turns, direct model streams, tool execution, and structured-session runs.\n\nMinimal configuration:\n\n```bash\nOTEL_ENABLED=true\nOTEL_SERVICE_NAME=swarmclaw\nOTEL_EXPORTER_OTLP_ENDPOINT=https://your-collector:4318\nOTEL_EXPORTER_OTLP_HEADERS=Authorization=Bearer your-token\n```\n\nIf you need a trace-specific endpoint, set `OTEL_EXPORTER_OTLP_TRACES_ENDPOINT` directly instead.\n\nOperational docs: https://swarmclaw.ai/docs/observability\n\n## Releases\n\n### v1.9.39 Highlights\n\nScheduled-run reliability release: stale workspace rebinding, fail-fast credential checks, and durable delivery evidence for schedule-driven sends.\n\n- **Legacy workspace cwd migration.** Scheduled reruns and reused schedule sessions pinned to a pre-migration workspace root (e.g. `~/.swarmclaw/workspace`) are now rebound to the current `WORKSPACE_DIR` automatically; intentional custom working directories are never touched.\n- **Orphan recovery dedup.** Startup queue recovery now recovers each orphaned task once instead of re-logging it every tick, and dead-letters tasks that repeatedly return to the queue without starting.\n- **Schedule delivery status.** Every scheduled run now writes `lastDeliveryStatus`/`lastDeliveryError` back to its schedule, so failures are visible without digging through task records.\n- **Credential preflight for schedules.** Scheduled runs on API-key providers fail fast with an actionable error when no credential resolves, instead of dying on a 401 deep in execution.\n- **Empty-run classification.** Runs that produce no text, no tool calls, and no error now fail with a clear provider-configuration message instead of a generic validation failure.\n- **Durable connector delivery evidence.** Task follow-up sends route through the connector outbox with task/schedule linkage, retries with backoff, and per-run dedupe, so triage can prove whether a scheduled send succeeded, failed, or was never attempted.\n- **Regression coverage.** Added tests for workspace path normalization, orphan recovery, schedule outcome writes, credential preflight, empty-run classification, and outbox-backed follow-ups.\n\n### v1.9.38 Highlights\n\nPR integration release for provider catalog coverage, OpenRouter context meters, and safer unsigned macOS desktop artifacts.\n\n- **TokenMix provider.** Added TokenMix as a built-in OpenAI-compatible provider with setup metadata, starter-agent defaults, and provider health checks.\n- **OpenRouter context meters.** Chat context status now uses cached OpenRouter model metadata when available so routed model context windows display accurately.\n- **macOS unsigned artifact fallback.** Desktop releases publish zip-only macOS artifacts when signing/notarization inputs are missing, avoiding the unsigned DMG damaged-app path.\n- **Regression coverage.** Added targeted tests for TokenMix setup, OpenRouter context metadata caching, and macOS target selection.\n\n### v1.9.37 Highlights\n\nTheme and memory-pressure release for lighter UI preferences and leaner chat history storage.\n\n- **Light, dark, and system theme modes.** Settings → Appearance now persists a Light/Dark/System selector while keeping the existing hue presets and custom color picker.\n- **Lean session history storage.** Legacy transcript blobs migrate into the `session_messages` table and are compacted from session records after persistence is verified, reducing page-load memory pressure on lower-RAM devices.\n- **Repo-backed message readers.** Global search, live usage summaries, and OpenClaw history merge now read table-backed messages after transcript compaction.\n- **Regression coverage.** Added tests for theme-mode normalization, legacy transcript compaction, and repo-backed message search.\n\n### v1.9.36 Highlights\n\nProtocol builder visibility release for built-in Structured Sessions.\n\n- **Built-in flow inspector.** Built-in protocol templates now open in a full-size visual builder canvas with a read-only template step panel.\n- **Canvas viewport repair.** Builder routes now claim the full dashboard workspace and refit React Flow after async template loads.\n- **Regression coverage.** Browser smoke now verifies that the built-in facilitated discussion graph renders with visible flow nodes.\n\n### v1.9.35 Highlights\n\nInstalled package build fix for fresh npm-global installs and upgrades.\n\n- **Fallback build dependency fix.** The npm package now declares `mime-types` and `@types/mime-types` directly so `swarmclaw server --build` can type-check the OpenClaw media proxy on clean installs.\n- **Installed-build regression guard.** CLI/package tests now verify that unbundled type declarations needed by local fallback builds ship as runtime package dependencies.\n- **macOS desktop status.** The damaged-app issue remains open until Developer ID signing and Apple notarization are configured and verified on downloaded macOS artifacts.\n\n### v1.9.34 Highlights\n\nCredential recovery and external extension access release for npm-global upgrades and scoped agent tool configuration.\n\n- **Credential secret recovery.** Startup now checks prior npm-global build env files before accepting a fresh per-version `CREDENTIAL_SECRET`, and validates candidate secrets against existing encrypted credentials before persisting `DATA_DIR/credential-secret`.\n- **Clear connector failures.** Connector startup now logs and surfaces credential decrypt failures directly instead of falling through to a misleading \"No bot token configured\" error.\n- **External extension tools.** Scoped agents now keep explicitly attached external `*.js` and `*.mjs` extensions, and the agent/chat tool controls persist enabled external tools through the `extensions` field.\n- **Regression coverage.** Added tests for previous-build credential recovery, non-decrypting secret replacement, scoped external extension access, and extension access persistence.\n\n### v1.9.33 Highlights\n\nIssue and PR validation release for credential durability, delegated task dispatch, connector output hygiene, and OpenClaw gateway protocol compatibility.\n\n- **Credential durability.** Execute-tool credential injection now reads the persisted `encryptedKey` field, and `CREDENTIAL_SECRET` now resolves in a stable order: explicit environment value, `DATA_DIR/credential-secret`, legacy env files, then generated fallback.\n- **Delegated task dispatch.** Agent-created tasks delegated to another agent auto-queue when no explicit status is supplied, and failed dead-lettered tasks can be requeued through `POST /api/tasks/:id/retry`.\n- **Connector output hygiene.** Connector replies now reuse the internal metadata scrubber before delivery and persistence, while successful non-connector delivery tool output is no longer overwritten as an unconfirmed send.\n- **Agent and gateway compatibility.** Agent updates preserve workspace filesystem settings, and OpenClaw gateway routes now use protocol version 4.\n- **Regression coverage.** Added tests for credential env injection, secret precedence, delegated queueing, failed-task retry, connector sanitization, agent workspace settings, and OpenClaw gateway protocol exports.\n\n### v1.9.32 Highlights\n\nPR integration release for background model routing, reflection memory controls, and current ClawHub install guidance.\n\n- **Background model routing.** Per-agent `dreamConfig` overrides can route dream cycles and daily digests before global dream settings, while `compactionProvider` settings can route live auto-compaction summaries through a cheaper or faster model.\n- **Reflection memory controls.** `reflectionMinQuality` gates automatic reflection memory writes without dropping the reflection record, and optional embedding dedup skips near-duplicate reflection notes when embeddings are configured.\n- **ClawHub install guidance.** OpenClaw skill docs now use `openclaw skills install swarmclaw` and current owner-scoped ClawHub links.\n- **Regression coverage.** Added tests for dream override precedence, compaction preference resolution, reflection quality gating, and embedding-based reflection dedup.\n\n### v1.9.31 Highlights\n\nDocumentation cleanup release for public release notes and OpenClaw guidance. No runtime behavior changed.\n\n- **Public docs cleanup.** Removed an unwanted third-party example from the README and site release notes.\n- **OpenClaw guidance preserved.** The README keeps the SwarmClaw-native OpenClaw gateway, skill, and agent-file guidance without naming unrelated workflows.\n\n### v1.9.30 Highlights\n\nPR integration release for dream-model routing, email bridge TLS opt-outs, and installed CLI runtime resolution.\n\n- **Dream model routing.** Memory dream cycles and daily digests can use optional `dreamProvider` settings so background consolidation can run on a smaller local model.\n- **Email bridge TLS opt-outs.** `tlsRejectUnauthorized=false` now disables hostname checks too, matching the explicit self-signed-server opt-out.\n- **Installed CLI stability.** Legacy API-backed CLI commands import the package-local `tsx` runtime instead of resolving `tsx` from the caller's project.\n\n### v1.9.29 Highlights\n\nIssue-fix release for Edit Agent tooltips, installed package builds, and structured dream output on local Ollama models.\n\n- **Edit Agent tooltips.** Help tips in the Edit Agent sheet now render above modal layers instead of being hidden behind the dialog.\n- **Installed package builds.** The npm package now ships the Dagre type declarations needed by `swarmclaw server --build`.\n- **Local Ollama dream output.** Structured dream/reflection calls request Ollama JSON mode and validate balanced JSON before writing memories.\n- **Regression coverage.** CLI/package, model-build, and dream-parser tests cover the reported failure modes.\n\n### v1.9.28 Highlights\n\nIssue-fix release for installed CLI groups, email bridge TLS handling, built-in model overrides, and Windows desktop native modules.\n\n- **Installed CLI groups.** Global npm installs route legacy API-backed group commands through the bundled TS runtime when installed under `node_modules`, avoiding Node 22.6+/25 type-stripping failures.\n- **Email bridge TLS resilience.** The email connector logs IMAP socket errors without crashing the daemon and supports `tlsRejectUnauthorized=false` for local self-signed IMAP/SMTP servers.\n- **Provider model override persistence.** Built-in provider live model saves now reload array-valued overrides instead of falling back to catalog defaults.\n- **Windows desktop native modules.** Desktop packaging syncs rebuilt Electron-native modules into traced `.next/node_modules` aliases so packaged Windows installs start against the correct ABI.\n- **Regression coverage.** CLI, email, provider route, and Electron after-pack tests cover the reported failure modes.\n\n### v1.9.27 Highlights\n\nDesktop compatibility and provider-save repair for Intel Mac users and OpenRouter setup.\n\n- **Intel macOS native modules.** The desktop packaging hook now rebuilds Electron-loaded native modules with the target architecture and blocks a release if an x64 macOS bundle contains an arm64-only required addon.\n- **OpenRouter save repair.** Provider updates now tolerate UI metadata fields like `id`, `type`, `createdAt`, and `updatedAt` without persisting them, while still rejecting unrelated unknown fields.\n- **Downloads clarity.** The downloads page no longer guesses Apple Silicon when a browser hides the Mac architecture, so Intel users can choose the x64 DMG explicitly.\n- **Regression coverage.** Provider route and Electron after-pack tests cover the reported failure modes.\n\n### v1.9.26 Highlights\n\nOutput hygiene follow-up: empty successful LLM turns now stay silent instead of being rewritten as user-visible errors.\n\n- **Silent empty completions.** Blank successful runs no longer become `Error: Run completed...` assistant messages.\n- **Connector-safe final text.** Slack and other connectors no longer receive synthetic error text for intentional silence or quiet no-op turns.\n- **Real errors preserved.** Explicit provider failures and streamed provider errors still surface as terminal errors.\n- **Regression coverage.** Chat-execution tests now lock the distinction between empty success and real failure.\n\n### v1.9.25 Highlights\n\nGateway lifecycle release: saved OpenClaw gateways now have explicit operator lifecycle controls, automatic routing avoids gateways that should not receive new work, and Slack peer-agent messages flow through the existing connector policy gates.\n\n- **Gateway lifecycle controls.** Providers can activate, drain, cordon, and request restart for saved OpenClaw gateway profiles.\n- **Routing guardrails.** OpenClaw route selection skips draining and cordoned profiles, including default, preferred, and pinned gateway paths.\n- **Operations Pulse awareness.** Cordoned and draining gateways now appear as operator attention items before they surprise a handoff or release check.\n- **CLI lifecycle access.** `swarmclaw gateways activate`, `drain`, `cordon`, and `restart` now post the matching lifecycle action for automation and release scripts.\n- **Slack peer collaboration.** Slack peer-bot messages are no longer dropped before group policy, mention, and self-loop protections run.\n\n### v1.9.23 Highlights\n\nSchedule reliability release: recurring work now repairs stale timing state before it can skip the nearest run, and scheduled board tasks keep mission context across repeat launches.\n\n- **Cron drift repair.** Active cron schedules repair missing or invalid `nextRunAt` values and stale future cron slots before the scheduler decides whether work is due.\n- **Tick-time advancement.** Cron and interval schedules now advance from the scheduler tick time instead of the process wall clock, making restart and catch-up behavior deterministic.\n- **Stable stagger.** Schedule stagger offsets are deterministic per schedule, avoiding thundering-herd launches without moving a saved next-run target on every recompute.\n- **Mission continuity.** Schedule-created board tasks attach to a persistent mission link, so recurring runs share the same operational context.\n\n### v1.9.22 Highlights\n\nResearch tools release: agents now get direct `web_extract` and `web_crawl` tools alongside `web_search`, `web_fetch`, and the unified `web` tool.\n\n- **Source-grounded extraction.** `web_extract` returns a page title, canonical URL, and readable content for known source URLs.\n- **Bounded crawls.** `web_crawl` walks same-origin links by default with conservative page and depth caps, plus an explicit external-link opt-in.\n- **Better routing.** Tool aliases, capability policy, planning hints, continuation recovery, and the chat UI all recognize the granular research tools.\n- **Regression coverage.** New tests cover action inference, tool-call translation, direct tool registration, extraction cleanup, and same-origin crawl bounds.\n\n### v1.9.21 Highlights\n\nProvider diagnostics release: connection checks now return a structured step timeline across setup, provider settings, and agent editing.\n\n- **Connection timelines.** Provider checks show endpoint resolution, model discovery, fallback selection, and chat/gateway verification steps.\n- **Safer error details.** Token-like values are redacted before check messages or diagnostics are returned to the UI.\n- **Local runtime debugging.** LM Studio, Ollama, custom OpenAI-compatible endpoints, cloud providers, OpenClaw gateways, and CLI providers all report concise pass/fail diagnostics.\n- **macOS signing path.** Desktop releases now forward Developer ID and Apple notarization credentials when configured, while ad-hoc fallback builds keep the quarantine workaround documented.\n\n### v1.9.20 Highlights\n\nProvider reliability release: local OpenAI-compatible runtimes now get safer endpoint handling, clearer setup, and first-class LM Studio support.\n\n- **LM Studio provider.** LM Studio is available in setup, provider settings, agent editing, model discovery, and connection checks with an optional API key.\n- **Endpoint normalization.** LM Studio and OpenAI-compatible OpenAI overrides normalize bare hosts like `http://127.0.0.1:1234` to `/v1` before calling models or chat completions.\n- **Provider switch isolation.** Switching an agent from a local endpoint back to a fixed cloud provider clears stale per-agent endpoints and fallback keys.\n- **Manual model flow.** Provider model saves now preserve explicit empty endpoint resets and optional-key providers can be tested without creating a credential.\n\n### v1.9.19 Highlights\n\nOutput hygiene release: final assistant responses now use the shared internal metadata scrubber before persistence, UI reset, connector delivery, and completion hooks.\n\n- **Multi-block scrubbing.** Repeated internal metadata payloads are stripped in one pass instead of stopping after the first block.\n- **Malformed prelude cleanup.** When a validated internal block is followed by a malformed internal fragment, the leftover prelude is removed before user-facing text is delivered.\n- **Shared finalizer path.** Post-stream finalization now uses the same metadata scrubber as the chat UI, keeping stored, streamed, and connector-visible output aligned.\n- **Regression coverage.** Tests cover repeated classifier-shape blocks, malformed follow-on fragments, and false-positive protection for malformed text without a prior validated strip.\n\n### v1.9.18 Highlights\n\nSchedule preflight release: schedules now show server-backed timing forecasts before save, with timezone-aware cron previews and warnings for risky drafts.\n\n- **Schedule preview API.** `POST /api/schedules/preview` validates a draft schedule through the same normalization path as saved schedules and returns the next calculated runs.\n- **Timezone-aware schedule sheet.** Cron schedules can set an explicit timezone, preview the next runs from the server, and see warnings before saving.\n- **Stagger and one-shot controls.** Operators can add a stagger window to recurring schedules and choose a run-once delay from the schedule workflow.\n- **CLI access.** `swarmclaw schedules preview --data '{...}'` exposes the same forecast for scripts and release automation.\n\n### v1.9.17 Highlights\n\nAgent configuration history release: SwarmClaw now surfaces saved agent versions directly in the agent editor, giving operators a fast rollback path for agent settings.\n\n- **Agent sheet history.** Advanced settings list recent saved versions with relative time, actor, and provider/model snapshot.\n- **One-click restore.** Operators can restore a prior agent configuration through the existing version-restore API without leaving the agent workflow.\n- **Stale-form protection.** Successful restore reloads agent state and closes the sheet so operators reopen the refreshed record.\n- **Regression coverage.** New tests cover config-version list/restore routes and summary formatting.\n\n### v1.9.16 Highlights\n\nAgent planning controls release: strict planning is now a first-class agent setting instead of a hidden persisted field, so operators can decide which agents must expose machine-readable plans before multi-step work.\n\n- **Agent editor control.** Advanced agent settings now include a Standard / Strict planning selector with inline behavior guidance.\n- **Runtime prompt wiring.** Strict planning continues to inject the existing `[MAIN_LOOP_PLAN]` contract before multi-step tool work, and the test suite now keeps that prompt section in the runtime gate.\n- **Portable agent packs.** Agent exports preserve `planningMode`, so planning discipline follows agents across installs.\n- **API coverage.** Agent create and update route tests verify that strict planning persists without clobbering unrelated settings.\n\n### v1.9.15 Highlights\n\nRun handoff release: SwarmClaw now turns completed, failed, queued, or running execution records into copyable handoff packets with outcome, evidence, artifacts, timeline, usage, resume commands, and recommended next actions.\n\n- **Run handoff API.** `GET /api/runs/:id/handoff` returns structured handoff JSON, and `?format=markdown` returns copyable markdown.\n- **Run Review copy action.** The run detail sheet exposes a copy handoff button so operators can move outcome evidence into another session without replaying the full event log.\n- **CLI access.** `swarmclaw runs handoff