AIエージェント
paritok-4b-v1
Compression gateway that cuts AI coding agents' token bills up to 85%.
★ 1.5k スターApache-2.0
ガイドを見る →babysitterは、a5c-aiが公開しているオープンソースのAIエージェントプロジェクトです（GitHub: a5c-ai/babysitter）。GitHubスターは1,837、フォークは113で、MITライセンスの下で公開されています。GitHubでの最終更新は2026年9月16日です。
Babysitter enforces the process: deterministic, quality-gated orchestration for fleets of coding agents.
Decide which coding harness you will run under Babysitter and how much human oversight (breakpoints) your workflow needs.
Install the main Babysitter CLI globally from npm, and optionally the adapters CLI and runtime platform packages.
Install the plugin for your harness, then run the user and project setup commands to create your profile and configure project workflows.
Start an orchestration run in interactive, planning, autonomous, or continuous mode and monitor it with the built-in commands.
Workflows defined in code are enforced step by step; progression requires gates to pass and approvals at configured breakpoints.
Every decision is recorded, giving an auditable trail for complex, long-running agent work.
The same processes run across 12 supported AI coding harnesses, avoiding lock-in to one vendor's agent tool.
Interactive mode pauses for approval, planning mode stops after design, autonomous (yolo) mode runs trusted tasks fully, and continuous mode runs indefinitely for monitoring-style work.
A built-in harness runs processes programmatically without any external coding agent, suiting CI/CD and scripted orchestration, and can delegate to discovered harnesses.
Doctor diagnostics, a live observation dashboard, resume for interrupted runs, retrospectives, and cleanup of old runs.
Most users install the published npm packages instead.
git clone https://github.com/a5c-ai/babysitter.git
cd babysitterInstall the main CLI globally, then add the plugin for your harness, for example Claude Code.
npm install -g @a5c-ai/babysitter
claude plugin marketplace add a5c-ai/babysitter-claude
claude plugin install --scope user babysitter@a5c.aiInside your harness, invoke the babysit skill on a real task.
claude "/babysitter:call implement user authentication with TDD"It is maintained by a5c-ai on GitHub.
Yes, Babysitter is MIT-licensed and free. The agents it orchestrates use your existing harness subscriptions or API keys.
Babysitter itself needs no key; it drives harnesses such as Claude Code or Codex, which authenticate with your own accounts.
Free-form agent chats improvise their process; Babysitter constrains agents to a coded workflow with gates, breakpoints, and a journal, trading flexibility for determinism and auditability.
npm install -g @a5c-ai/babysitter, install your harness plugin, run the user and project setup commands, then start a run with /babysitter:call.
Compression gateway that cuts AI coding agents' token bills up to 85%.
Governance and orchestration for AI agent fleets: declarative rules, verifiable runs.
Hook and skill guard that stops coding agents from doing unrequested busywork.
Manage parallel AI coding agents on an infinite tmux-backed canvas.
A markup-driven CSS orchestration language for developers and AI agents.
Dev agent multiplexer designed for git worktrees and coding agents.
LibreAIカタログに掲載されたAIエージェントプロジェクトの中で最もスターの多いもの — babysitterが合わなかった場合に検討したい、現実的な候補リストです。
A personal AI that really does things — on any OS, with your data staying yours.
An agentic skills framework & software development methodology that works.
Agent harness optimization system with skills, memory and security for Claude Code, Codex and more
The agent that grows with you — a Python framework for adaptive AI agents.
DeepSeek's plugin-based agent harness where everything is a plugin.
カタログの中でbabysitterとスター数が近いAIエージェントプロジェクト — 前後の棚もそのまま見て回れます。
Go framework for production agent systems: graphs, tools, memory, A2A, MCP.
Management console for installing and monitoring AI agents and pipelines.
Gortex, in Go — built for high-performance code-intelligence engine for ai agents.
Android GUI agent framework: AI that sees, plans, and taps real mobile apps.
Local-first AI chat workspace for models.
Self-hosted ai agents: minimal ai coding agent (~1,000 lines of python) inspir.