Babysitter orchestrated-execution

Execute work units through the rigorous 4-phase Metaswarm cycle (Implement -> Validate -> Adversarial Review -> Commit) with independent quality gate enforcement.

install

source · Clone the upstream repo

git clone https://github.com/a5c-ai/babysitter

Claude Code · Install into ~/.claude/skills/

T=$(mktemp -d) && git clone --depth=1 https://github.com/a5c-ai/babysitter "$T" && mkdir -p ~/.claude/skills && cp -r "$T/library/methodologies/metaswarm/skills/orchestrated-execution" ~/.claude/skills/a5c-ai-babysitter-orchestrated-execution && rm -rf "$T"

manifest: library/methodologies/metaswarm/skills/orchestrated-execution/SKILL.md

Orchestrated Execution

Overview

Execute work units through the Metaswarm 4-phase cycle. Quality gates are blocking state transitions. Adversarial reviews use fresh reviewers with binary PASS/FAIL verdicts.

When to Use

Implementing work units from an approved plan
Executing tasks requiring rigorous quality enforcement
When TDD compliance must be verified independently

Process

IMPLEMENT - Coder executes via TDD against spec with DoD items
VALIDATE - Orchestrator independently runs tsc, eslint, vitest (never trusts subagent)
ADVERSARIAL REVIEW - Fresh reviewer checks spec compliance: binary PASS/FAIL with evidence
COMMIT - Only after adversarial PASS, within declared file scope

Anti-Patterns (Enforced)

Self-certifying (trusting subagent "tests passed" claims)
Combining phases into single steps
Reusing reviewers after FAIL
Passing previous review findings to new reviewers
Treating quality gate failures as advisory

Agents Used

```
agents/coder/
```
- TDD implementation
```
agents/code-reviewer/
```
- Fresh adversarial review
```
agents/issue-orchestrator/
```
- Independent validation

Tool Use

Invoke via babysitter process:

methodologies/metaswarm/metaswarm-execution-loop