a5c-ai/babysitter

Babysitter enforces obedience on agentic workforces and enables them to manage extremely complex tasks and workflows through deterministic, hallucination-free self-orchestration

What it solves

Babysitter prevents AI agents from hallucinating or deviating from intended workflows. It provides a deterministic orchestration layer that ensures agentic workforces follow a strictly defined process, requiring human approval at specific breakpoints and enforcing quality gates before moving to the next step.

How it works

Workflows are defined as JavaScript code rather than simple configuration files, making the code the sole authority for the agent's behavior. The system uses a runtime called Adapters, which allows the same process definitions to run across 12 different AI coding harnesses (such as Claude Code, Codex, and Cursor). It records every decision in an immutable journal to maintain a complete audit trail.

Who it’s for

Developers and teams using AI coding agents who need high-reliability, complex workflows that cannot afford the unpredictability of fully autonomous agents.

Highlights

  • Harness-Agnostic: Works across a wide variety of AI coding tools including Claude Code, Codex, Cursor, and GitHub Copilot.
  • Deterministic Orchestration: Uses code-based process definitions to eliminate hallucinations and ensure obedience.
  • Human-in-the-Loop: Enforces mandatory breakpoints for human approval.
  • Internal Harness: Includes a built-in engine for headless orchestration in CI/CD pipelines without requiring an external AI agent.

Related

  • Project
    sipyourdrink-ltd/bernstein

    An open-source governance layer for AI agents that uses deterministic scheduling and policy-as-code to ensure agent runs are isolated, reproducible, and verifiable via signed audit receipts.

  • Project
    Arch1eSUN/Arcgentic

    A harness engineering layer for Codex and Claude Code that turns AI coding into a gated workflow with specialized roles for planning, development, and auditing.

  • Project
    appautomaton/latex-arxiv-SKILL

    An agentic harness for writing verified AI/ML review papers in LaTeX, featuring a gated, issue-driven workflow that prevents citation hallucinations.

  • Project
    affaan-m/ECC

    An agent harness operating system that provides AI agents with a coordinated engineering system, including 68 agents and 286 skills to implement professional software engineering workflows.

  • Project
    Chorus-AIDLC/Chorus

    Chorus is a management harness for AI coding agents that implements the AI-DLC workflow, allowing teams to orchestrate multiple agents and ensure human verification of AI-proposed code.