Help us improve
Share bugs, ideas, or general feedback.
Share bugs, ideas, or general feedback.
Share bugs, ideas, or general feedback.
By a5c-ai
Orchestrate complex, multi-step workflows with event-sourced state management, hook-based extensibility, and human-in-the-loop approval
npx claudepluginhub a5c-ai/babysitter-claude --plugin babysitterAssimilate an external methodology, harness, or specification into babysitter process definitions with skills and agents.
---
Clean up .a5c/runs and .a5c/processes directories. Aggregates insights from completed/failed runs into docs/run-history-insights.md, then removes old run data and orphaned process files.
---
Diagnose babysitter run health - journal integrity, state cache, effects, locks, sessions, logs, and disk usage
Matches all tools
Hooks run on every tool call, not just specific ones
Executes bash commands
Hook triggers when Bash tool is used
Share bugs, ideas, or general feedback.
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge.
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge.
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Harness-native ECC plugin for engineering teams - 64 agents, 262 skills, 84 legacy command shims, reusable hooks, rules, MCP conventions, and operator workflows for Claude Code plus adjacent agent harnesses
Core skills library for Claude Code: TDD, debugging, collaboration patterns, and proven techniques
Tools to maintain and improve CLAUDE.md files - audit quality, capture session learnings, and keep project memory current.
Manus-style persistent markdown files for planning, progress tracking, and knowledge storage. Works with Claude Code, Kiro, Clawd CLI, Gemini CLI, Cursor, Continue, Hermes, and 17+ AI coding assistants. Now with Arabic, German, Spanish, and Chinese (Simplified & Traditional) support.
v9.44.0 — Patch release for cursor-agent smoke checks in untrusted workspaces. Run /octo:setup.
Unity Development Toolkit - Expert agents for scripting/refactoring/optimization, script templates, and Agent Skills for Unity C# development
Implementation of the babysitter technique - continuous orchestration loops for deterministic development. Run Claude in a loop with orchestration steps based on the babysitter-sdk and technique.
Turn a stated need into a full system design by mining the Atlas knowledge graph.
Babysitter is the Claude Code plugin for SDK-backed orchestration with event-sourced runs, stop-hook continuation, native hook dispatch, and human approval gates.
Status: Public harness plugin surface for Claude Code. Canonical docs home: Package and Plugin Docs Map. This README is the canonical Claude Code plugin contract; use Plugins Overview for repo-level plugin discovery.
Claude Code is the canonical hook-driven harness:
SessionStart prepares baseline session statecompletionProof is emitted and returned as
<promise>...</promise>Do not document multi-iteration loops inside a single turn.
Process discovery should prefer:
.a5c/processes in the current workspacebabysitter process-library:active --jsonThe process library is active-use and layered. Docs should not frame it as a single bundled snapshot or as a team/project-scoped library only.
Current generated-process guidance:
agentskill when a matching installed skill existsshell for existing tooling/tests/buildsbreakpoint for approval gatessleep for time gatesDo not document node as an allowed generated effect kind for current
Babysitter processes.
The Claude-facing README should stay on the Claude skill and hook experience. Low-level Babysitter runtime mechanics such as run creation, result posting, binding, and compatibility fallbacks belong in the babysit skill and hook implementation docs, not in this top-level README.
Interactive breakpoint rules:
--status ok and approved: falseThe plugin dispatches native Babysitter hooks from these priority tiers:
.a5c/hooks/<hook-type>/~/.config/babysitter/hooks/<hook-type>/plugins/babysitter/hooks/<hook-type>/Common hook points:
on-run-starton-iteration-starton-iteration-endon-task-starton-task-completeon-breakpointon-run-completeon-run-failplugins/babysitter/plugin.jsonplugins/babysitter/hooks/plugins/babysitter/skills/babysit/SKILL.mdlibrary/Install the Claude Code plugin through the Claude plugin installer or the SDK CLI helper:
claude plugin marketplace add a5c-ai/babysitter
claude plugin install --scope user babysitter@a5c.ai
The SDK helper runs the same published Claude plugin flow:
babysitter harness:install-plugin claude-code
Then restart Claude Code. The installer does not materialize a repo-local plugin copy.
Claude Cowork can install this same plugin package without any plugin-structure
changes. This repository already includes a Claude marketplace manifest at
.claude-plugin/marketplace.json that points at plugins/babysitter/.
Cowork tab.Customize in the left sidebar.Browse plugins.Personal.+, then choose Add marketplace from GitHub.https://github.com/a5c-ai/babysitter.Babysitter from the marketplace that Cowork imports from this repo.After installation, the plugin's skills appear in Cowork when you type / or
click +.
Organization owners can also distribute Cowork plugins from
Organization settings > Plugins.
owner/repo format.Anthropic's current Cowork docs require organization GitHub-synced
marketplaces to use private or internal repositories on github.com, not
public repos. If you want to distribute Babysitter org-wide through a
GitHub-synced marketplace, use a private/internal fork of this repo or upload
the plugin ZIP manually.
If the workspace does not already have an active process-library binding, this command bootstraps the shared global SDK process library automatically:
babysitter process-library:active --json