# Clawboo

## Docs

- [Changelog](https://docs.claw.boo/appendices/changelog.md): Release history of the clawboo CLI from the 0.1.0 name-claim placeholder through v0.2.0, the liberated cut.
- [Contributing to Clawboo](https://docs.claw.boo/appendices/contributing.md): Clone, run the dev stack, pass the local gate, add a changeset, and respect the architecture invariants.
- [Clawboo Frequently Asked Questions: Runtimes, Data, and Cost](https://docs.claw.boo/appendices/faq.md): Answers to common Clawboo questions — runtimes, data storage, cost, network exposure, unattended agents, resetting, Atlas versus the Ghost Graph.
- [Glossary](https://docs.claw.boo/appendices/glossary.md): Canonical definitions of Clawboo's core terms, each linked to the page that explains it.
- [Known issues & limitations](https://docs.claw.boo/appendices/known-issues.md): Code-verified list of Clawboo's genuine current limitations, deferrals, and dormant seams, not fixed bugs.
- [License & third-party notices](https://docs.claw.boo/appendices/license-and-notices.md): Clawboo is MIT-licensed. Summary of the bundled runtime stack, their licenses, and the MIT-licensed upstream marketplace content.
- [The agent model](https://docs.claw.boo/concepts/agent-model.md): What a Boo is, how Boo Zero leads, and how a runtime-agnostic agent record maps onto the five runtime classes.
- [Capabilities](https://docs.claw.boo/concepts/capabilities.md): The unified capability inventory, one CapabilityRecord stream fanned from five per-runtime sources, with manageability tiers that decide what Clawboo can change.
- [Delegation and orchestration](https://docs.claw.boo/concepts/delegation-and-orchestration.md): How Clawboo turns structured delegation signals from runtime events into durable board mutations instead of scraping prose.
- [Gateway and events](https://docs.claw.boo/concepts/gateway-and-events.md): The OpenClaw Gateway flow through the same-origin WS proxy, and the Bridge→Policy→Handler event pipeline that turns raw frames into UI state.
- [Governance](https://docs.claw.boo/concepts/governance.md): USD budgets with an auto-pause kill-switch, tool-loop circuit breakers, depth/fan-out/cost caps, and the delegation approval handshake.
- [Core concepts](https://docs.claw.boo/concepts/index.md): The concept map for Clawboo's orchestration model, agents, teams, the board, delegation, memory, governance, and the invariants that hold them together.
- [Memory in Clawboo: Shared and Private Tiers Explained](https://docs.claw.boo/concepts/memory.md): How Clawboo's shared Memory-MCP store lets every runtime save and recall team knowledge, while each runtime's native private memory stays preserved.
- [Observability](https://docs.claw.boo/concepts/observability.md): Clawboo's always-on orchestration event log, the projections it feeds, the error taxonomy, and the lazy OpenTelemetry bridge.
- [Peer chat](https://docs.claw.boo/concepts/peer-chat.md): The mixed-runtime team room where every runtime is a named peer, with bounded speaker selection and isUser=false evidence tagging.
- [Scheduling Recurring Team Work with Clawboo Routines](https://docs.claw.boo/concepts/scheduling.md): Create Routines that fire recurring or one-shot board tasks on cron schedules, survive restarts, respect budgets and verification, and halt on failure.
- [Teams and planes](https://docs.claw.boo/concepts/teams-and-planes.md): How teams group agents, and the split between the shared coordination plane and each runtime's private plane.
- [The Clawboo Board: Durable Kanban Task Coordination](https://docs.claw.boo/concepts/the-board.md): How Clawboo's seven-column board tracks work atomically, recovers from crashes, chains task dependencies, and narrates decisions into team chat.
- [Verification in Clawboo: Builder ≠ Judge Principle](https://docs.claw.boo/concepts/verification.md): How Clawboo's two-layer verification gate — a deterministic build check and an independent read-only critic — makes `done` mean verified work.
- [Worktrees and handoff](https://docs.claw.boo/concepts/worktrees-and-handoff.md): The per-task git-worktree system-of-record and the runtime-agnostic AGENT_HANDOFF.json that lets one runtime pick up another's work.
- [Dashboard tour](https://docs.claw.boo/getting-started/dashboard-tour.md): A map of the Clawboo dashboard, the three-column shell, the view modes, Atlas vs the per-team Ghost Graph, and every nav panel.
- [Deploy your first team](https://docs.claw.boo/getting-started/first-team.md): Deploy a team of agents, run the Know-Your-Team gate, send a message in group chat, and watch a delegation land on the board.
- [Install Clawboo](https://docs.claw.boo/getting-started/installation.md): Run npx clawboo to launch the dashboard. Prerequisites, what the launcher does, port discovery, and where Clawboo keeps its state.
- [Native runtime quickstart: your first Clawboo agent team](https://docs.claw.boo/getting-started/quickstart-native.md): Paste one API key from Anthropic, OpenAI, OpenRouter, or Ollama and get a working two-agent team running in under 60 seconds — no Gateway required.
- [Quickstart: deploy a team over the OpenClaw Gateway](https://docs.claw.boo/getting-started/quickstart-openclaw.md): Detect, install, and configure the OpenClaw Gateway from inside Clawboo's wizard, approve device pairing, then deploy a team over the live Gateway.
- [Attach an external agent to Clawboo's four MCP servers](https://docs.claw.boo/guides/attach-mcp-external-agent.md): Connect a standalone Claude Code, Codex CLI, or custom MCP client to Clawboo's Tasks, Memory, Tools, and TeamChat servers to join a team's shared board.
- [Cross-runtime handoff](https://docs.claw.boo/guides/cross-runtime-handoff.md): Pause a file-mutating board task on one runtime and resume it cold on another, using the per-task worktree and AGENT_HANDOFF.json.
- [Configure governance, budgets, and caps in Clawboo](https://docs.claw.boo/guides/governance-and-budgets.md): Add a USD hard-cap budget to a team, trigger the auto-pause kill-switch, understand the five circuit breakers, and resolve risky-delegation approvals.
- [Guides (cookbook)](https://docs.claw.boo/guides/index.md): Task-oriented walkthroughs that compose Clawboo features into end-to-end workflows.
- [Build a multi-runtime team](https://docs.claw.boo/guides/multi-runtime-team.md): Connect two or three runtimes, put agents from each on one team, and dispatch work that flows across them via the shared board, chat, and memory.
- [Schedule and manage recurring team work in Clawboo](https://docs.claw.boo/guides/recurring-team-work.md): Create a Routine with a cron expression, assign it to a team, test it with Run Now, and manage the error-halts policy so scheduled work stays reliable.
- [Self-host Clawboo securely](https://docs.claw.boo/guides/self-host-securely.md): Keep the loopback default, or widen the bind safely with an access token, a reverse proxy, and TLS, with the honest single-tenant caveats.
- [AgentSource (registry of record)](https://docs.claw.boo/internals/agent-source.md): The AgentSource trait, the neutral record types, the registry multiplexer, and the two concrete sources that make SQLite the agent registry of record.
- [Codegen and ingestion](https://docs.claw.boo/internals/codegen-and-ingestion.md): How the marketplace catalog is generated from pinned upstream repos, the zero-loss identityTemplate invariant, and the verify:ingest drift gate that keeps committed codegen honest.
- [Design system](https://docs.claw.boo/internals/design-system.md): How Clawboo's theme, surface tiers, motion, type scale, brand palette, and reusable UI primitives fit together, grounded in globals.css.
- [Event pipeline internals](https://docs.claw.boo/internals/event-pipeline.md): How @clawboo/events turns raw OpenClaw Gateway frames into store updates through pure Bridge and Policy layers and a stateful Handler with a RAF patch queue.
- [Monorepo and build](https://docs.claw.boo/internals/monorepo-and-build.md): How Clawboo's TurboRepo + pnpm-workspaces monorepo is laid out, the build order, what publishes, and the root commands.
- [The CapabilitySource and ScheduleSource seams](https://docs.claw.boo/internals/seams.md): The two read-fan-out multiplexers, CapabilitySource and ScheduleSource, that mirror AgentSource and merge per-source adapters into one normalized record stream.
- [Clawboo documentation](https://docs.claw.boo/intro/index.md): Clawboo is an open-source, local-first multi-agent mission-control dashboard and orchestrator you install with npx clawboo.
- [What is Clawboo](https://docs.claw.boo/intro/what-is-clawboo.md): A TypeScript orchestrator for heterogeneous AI agent runtimes, the three design wedges that distinguish it from running runtimes standalone.
- [Data and state](https://docs.claw.boo/operating/data-and-state.md): Where Clawboo stores everything, how to back it up, and how the hard-reset model works.
- [Deploying Clawboo](https://docs.claw.boo/operating/deployment.md): Run Clawboo via npx, in dev mode, or as a bundled server, covering ports, state dir, host binding, and reverse-proxy setup.
- [Attaching Clawboo's MCP servers](https://docs.claw.boo/operating/mcp-servers.md): Attach the four hosted MCP servers (tasks, memory, tools, teamchat) to an external agent over stdio or Streamable HTTP.
- [Security model and safe exposure](https://docs.claw.boo/operating/security.md): How Clawboo authenticates, isolates secrets, and redacts output, and how to expose the dashboard safely.
- [CLI reference](https://docs.claw.boo/reference/cli.md): The single npx clawboo launcher (only --version/--help) plus the four bundled MCP stdio bins.
- [Configuration reference](https://docs.claw.boo/reference/configuration.md): settings.json schema, the ~/.clawboo state directory, every file Clawboo owns, and OpenClaw interop reads.
- [Database schema reference](https://docs.claw.boo/reference/database-schema.md): The 27 SQLite tables Clawboo persists, their columns, indexes, and how the schema is bootstrapped without a migration ladder.
- [Clawboo Environment Variables: Complete Reference Guide](https://docs.claw.boo/reference/environment-variables.md): Every environment variable Clawboo reads — state paths, port binding, access tokens, provider keys, operational tuning, and OpenTelemetry export.
- [Orchestration events and error taxonomy](https://docs.claw.boo/reference/events-and-errors.md): Reference for the append-only orchestration event kinds and the runtime error classification used across observability.
- [MCP tools reference](https://docs.claw.boo/reference/mcp-tools.md): The four clawboo-hosted MCP servers (tasks, memory, tools, teamchat) and every tool each exposes, with its zod input schema.
- [@clawboo/adapter-claude-code](https://docs.claw.boo/reference/packages/adapter-claude-code.md): Claude Code RuntimeAdapter that maps the Claude Agent SDK lifecycle into the normalized executor event stream.
- [@clawboo/adapter-codex](https://docs.claw.boo/reference/packages/adapter-codex.md): Codex RuntimeAdapter that maps the codex exec event stream into the normalized executor event stream.
- [@clawboo/adapter-hermes](https://docs.claw.boo/reference/packages/adapter-hermes.md): Hermes RuntimeAdapter that drives the Hermes CLI as a board worker, mapped into the normalized executor event stream.
- [@clawboo/adapter-native](https://docs.claw.boo/reference/packages/adapter-native.md): The pure RuntimeAdapter for clawboo's in-process native conversational harness, plus its AgentConfig schema and native→RuntimeEvent mapper.
- [@clawboo/adapter-openclaw](https://docs.claw.boo/reference/packages/adapter-openclaw.md): OpenClaw RuntimeAdapter that wraps the Gateway client into the normalized executor lifecycle stream.
- [@clawboo/agent-registry](https://docs.claw.boo/reference/packages/agent-registry.md): The AgentSource trait, neutral agent/team/session record types, and the AgentRegistry multiplexer, the agent-registry-of-record seam.
- [@clawboo/boo-avatar](https://docs.claw.boo/reference/packages/boo-avatar.md): Deterministic ghost-lobster SVG avatar generator for Clawboo agents. Pure, browser-safe, zero-dep.
- [@clawboo/capability-registry](https://docs.claw.boo/reference/packages/capability-registry.md): Neutral CapabilityRecord types plus the CapabilitySource trait and read-fan-in multiplexer for the unified capability inventory.
- [@clawboo/events](https://docs.claw.boo/reference/packages/events.md): The Bridge → Policy → Handler event pipeline that turns raw Gateway frames into Zustand store mutations.
- [@clawboo/executor](https://docs.claw.boo/reference/packages/executor.md): The RuntimeAdapter trait, the normalized RuntimeEvent lifecycle union, the adapter contract suite, and KV-cache prompt-tier primitives.
- [@clawboo/gateway-proxy](https://docs.claw.boo/reference/packages/gateway-proxy.md): Same-origin WebSocket proxy for the OpenClaw Gateway with server-side token + Ed25519 device-auth injection.
- [@clawboo/logger](https://docs.claw.boo/reference/packages/logger.md): Pino structured logger plus the display/log-layer credential redactor shared across Clawboo packages and the Express server.
- [@clawboo/mcp](https://docs.claw.boo/reference/packages/mcp.md): Protocol adapters exposing Clawboo's Tasks / Memory / Tools / TeamChat MCP servers over stdio and in-process Streamable HTTP.
- [@clawboo/protocol](https://docs.claw.boo/reference/packages/protocol.md): Message parser, transcript types, and agent file definitions for Clawboo. Pure, browser-safe, zero-dep.
- [@clawboo/tsconfig](https://docs.claw.boo/reference/packages/tsconfig.md): Shared TypeScript compiler-option presets (base, react, node) every Clawboo package and app extends via the tsconfig "extends" field.
- [@clawboo/worktrees](https://docs.claw.boo/reference/packages/worktrees.md): Server-only per-task git-worktree lifecycle, system-of-record scaffold, and structured cross-runtime AGENT_HANDOFF.json bridge.
- [Clawboo Agents API: Create, Read, Update, Delete Boos](https://docs.claw.boo/reference/rest-api/agents.md): Full REST reference for the Clawboo agent registry — list, create, sync, read, delete, manage agent files, and list live sessions for each Boo.
- [Clawboo Board API: Task Lifecycle and Atomic Claim](https://docs.claw.boo/reference/rest-api/board.md): REST reference for the Clawboo durable board — create and claim tasks, transition status, record executions, link dependencies, and manage worktrees.
- [Clawboo Governance API: Budgets, Audit, and Approvals](https://docs.claw.boo/reference/rest-api/governance.md): REST reference for Clawboo governance — list and set USD budgets, resume a paused scope, read the forensic audit log, and route delegation approvals.
- [REST API overview](https://docs.claw.boo/reference/rest-api/index.md): Base URL, the SPA proxy, the access gate, the WS upgrade, the error envelope, and a grouped index of all 124 routes.
- [Clawboo Memory API: Read and Write Shared Agent Memory](https://docs.claw.boo/reference/rest-api/memory.md): REST reference for Clawboo's shared memory tier — search facts, save facts and procedures, browse stored entries, and inspect the embedding provider.
- [Misc resources API](https://docs.claw.boo/reference/rest-api/misc.md): REST reference for cost records, chat history, graph layout, personality, skills, exec settings, fleet summary, and Boo Zero context.
- [Clawboo Observability API: Event Log and Fleet Health](https://docs.claw.boo/reference/rest-api/observability.md): REST reference for Clawboo observability — query the event log, reconstruct traces, read fleet health, stream live events over SSE, and run eval smoke.
- [Clawboo Runtimes API: Connect, Install, and Run Agents](https://docs.claw.boo/reference/rest-api/runtimes.md): REST reference for Clawboo's four runtimes — list capabilities, install CLI tools, connect provider keys, verify keys, drive tasks, seed a native team.
- [Clawboo Schedules API: Routines and Gateway Cron Reference](https://docs.claw.boo/reference/rest-api/schedules.md): REST reference for Clawboo's unified scheduler — list, create, patch, pause, resume, delete, and force-run schedules across Routines and Gateway cron.
- [Settings & Health API](https://docs.claw.boo/reference/rest-api/settings.md): REST reference for /api/settings (gateway URL + token) and /api/health (the boot probe / System Health surface).
- [Clawboo Teams API: Create, Manage, and Chat with Teams](https://docs.claw.boo/reference/rest-api/teams.md): REST reference for Clawboo teams — CRUD, agent membership, onboarding flags, team rules, and the peer-chat room including exchange kickoff.
- [Tools & MCP API](https://docs.claw.boo/reference/rest-api/tools-and-mcp.md): REST reference for the tools broker (list, approvals, audit) and the in-process Streamable-HTTP MCP transport plus attach-config endpoint.
- [Claude Code Runtime: Anthropic Agent SDK for Clawboo](https://docs.claw.boo/runtimes/claude-code.md): Use Anthropic's Claude Code agent SDK on your Clawboo team — 200k-token context, isolated git worktrees, streaming output, and real USD cost per run.
- [Codex Runtime: OpenAI Codex Exec Agent for Clawboo](https://docs.claw.boo/runtimes/codex.md): Run OpenAI's codex exec on your Clawboo team — authenticate via ChatGPT OAuth, spawn per task in an ephemeral home, and stream results to the board.
- [Connecting runtimes](https://docs.claw.boo/runtimes/connecting-runtimes.md): Install, connect, disconnect, and health-check the non-OpenClaw runtimes from the Runtimes panel.
- [Hermes runtime](https://docs.claw.boo/runtimes/hermes.md): Connect and run the Hermes CLI as a one-shot worker on Clawboo's board with a persistent, compounding per-identity home.
- [Runtimes overview](https://docs.claw.boo/runtimes/index.md): The five runtimes Clawboo can drive, the three runtime classes, and the full capability matrix that decides how each one is integrated.
- [Clawboo Native Runtime: Built-In, No-Gateway Agent Host](https://docs.claw.boo/runtimes/native.md): Run AI agents with no extra installs. Clawboo Native connects directly to Anthropic, OpenAI, OpenRouter, or a local Ollama instance using one provider key.
- [OpenClaw](https://docs.claw.boo/runtimes/openclaw.md): Connect the OpenClaw runtime, a connected substrate driven over the Gateway, with device pairing, registry sync, and global memory scope.
- [Working with agents](https://docs.claw.boo/using/agents.md): Open the agent detail view, edit the four agent files, tune the personality sliders, and create or delete a Boo.
- [Review and resolve approvals](https://docs.claw.boo/using/approvals.md): Use the Approvals panel to allow or deny pending exec approvals and the shared tool/delegation queue, team-scoped.
- [Track Every Agent Task on the Clawboo Kanban Board](https://docs.claw.boo/using/board.md): Track every agent task across seven kanban columns, read verification verdicts, inspect execution history, and drill into the task-detail drawer.
- [Configuring Boo Zero](https://docs.claw.boo/using/boo-zero.md): Set Boo Zero's display name, global brief, and per-team brief and rules, the universal team leader's identity and behavior.
- [Capabilities dashboard](https://docs.claw.boo/using/capabilities-dashboard.md): Use the Capabilities dashboard to browse the unified capability inventory and enable or disable the rows your runtimes let you manage.
- [Cost and budgets](https://docs.claw.boo/using/cost-and-budgets.md): Track token usage on the Cost dashboard and set USD budgets (warn or cap) per agent, mission, or team from Governance.
- [Visualize Your Agent Fleet with Ghost Graph & Atlas](https://docs.claw.boo/using/ghost-graph.md): Read your org chart at a glance — see active Boos, expand skill orbitals, draw routing edges, and switch between per-team and all-teams Atlas views.
- [Use the Governance dashboard](https://docs.claw.boo/using/governance-dashboard.md): Set budgets, raise caps, resume paused scopes, review the enforced-in-code caps, the approval queue, and the forensic audit log.
- [Using group chat](https://docs.claw.boo/using/group-chat.md): Open a team's Group Chat, pass the Know-Your-Team gate, send messages, route with @mentions, and watch delegations land on the board.
- [Browse and deploy from the Marketplace](https://docs.claw.boo/using/marketplace.md): Search, inspect, and deploy single agents or whole teams from Clawboo's 304-agent / 82-team / 30-skill catalog.
- [Browse and search shared memory](https://docs.claw.boo/using/memory-browser.md): Use the Memory panel to search, save, and browse the facts and procedures your team's runtimes share.
- [Using the Observability dashboard](https://docs.claw.boo/using/observability-dashboard.md): Read traces, error taxonomy, fleet health, the delegation graph, and run the smoke evals from the Observability panel.
- [Schedule recurring team work](https://docs.claw.boo/using/scheduler.md): Use the Scheduler tab to create, pause, run, and delete Routines and OpenClaw cron from one unified surface.
- [Check System Health](https://docs.claw.boo/using/system-health.md): Read the boot-probe checklist, the degraded/fatal banner, resolved runtime state, and re-run the probe from the System Health panel.
- [Create and manage teams](https://docs.claw.boo/using/teams.md): Create a team from a marketplace template or empty, set leaders and rules, pick color collections, and archive or delete it.
- [Theming](https://docs.claw.boo/using/theming.md): Switch Clawboo between light, dark, and system theme, and how the preference is resolved, persisted, and applied without a flash.