AI Controls Author — Testing & QA agent for Claude Code
AI system spec, eval suite, tool permission matrix, HITL workflow (taxonomy 4.15).
How to install AI Controls Author
Installs to ~/.claude/agents/lebobo88-pair-programmer-ai-controls-author.md
mkdir -p ~/.claude/agents && curl -fsSL https://raw.githubusercontent.com/lebobo88/pair-programmer/HEAD/.claude/agents/ai-controls-author.md -o ~/.claude/agents/lebobo88-pair-programmer-ai-controls-author.md Restart Claude Code, or start a new session, for it to be picked up.
What AI Controls Author does
name: ai-controls-author model: claude-opus-5 description: AI system spec, eval suite, tool permission matrix, HITL workflow (taxonomy 4.15). Used by ai-controls-team. tools: Read, Write, Edit, Glob, Grep, mcp__pp_harness__archive_artifact, mcp__pp_harness__record_attempt color: blue
You produce AI-feature governance artifacts. Judge applies `nist-ai-rmf-govern@1` and `nist-ai-rmf-measure@1`.
Stage kinds
ai_system_spec: model selection rationale, prompt boundaries, tool boundar
Alternatives in Testing & QA
- Eval Author — Authoring role for the D365FO agent eval loop catalog 136 ★
- Test Matrix Keeper — Expand or review the verification matrix for a growing harness surface 58 ★
- AI Provider Author — WRITE-access implementer for AI provider integrations, model routing, embeddings, prompt systems, streaming, a 52 ★
Full documentation available on GitHub
View Source RepositoryRelated Agents
Issue Team PM
PM agent for issue-team skill. Feature-only. Writes the spec, waits for team-lead approval, briefs Dev and QA,
Team Merger
Agent-teams merger. Merges each reviewed-and-approved worktree into the base branch in turn, resolves conflict
Yalla Tester
Test writer and runner for Yalla Coding Team. Writes tests, runs test suite, validates behavior. Does NOT writ
Regression Test Author
Writes a test that reproduces a bug BEFORE the fix is written. Invoke for every bug. The reproduction test sta
Test Author
Writes an executable test suite from the requirement and the existing public API only; never reads the impleme
Fixme Handle Code Review
Triages code review findings using the unified taxonomy plus edge-case validity classifications (FIX_FAIL_FAST
Related Skills
AI Gate
AI verification for G4 — eval suite against baseline, red-team status, guardrail verification, drift check
Eval Suite
Re-score tiny-spec against the SDD evaluation rubric, record it, and report the delta vs the last run.
Enterprise Incident Agent Platform
Production-oriented agentic incident-response platform using Claude on Amazon Bedrock, structured tool calling