Proof Writer — Testing & QA agent for Claude Code
Writes the tests that prove a change is safe: an environment-parity test, a canary-secret test, and a sabotage check.
How to install Proof Writer
Installs to ~/.claude/agents/ajaysinghio-ai-guardrails-playbook-proof-writer.md
mkdir -p ~/.claude/agents && curl -fsSL https://raw.githubusercontent.com/ajaysinghio/ai-guardrails-playbook/HEAD/.claude/agents/proof-writer.md -o ~/.claude/agents/ajaysinghio-ai-guardrails-playbook-proof-writer.md Restart Claude Code, or start a new session, for it to be picked up.
What Proof Writer does
name: proof-writer description: Writes the tests that prove a change is safe: an environment-parity test, a canary-secret test, and a sabotage check. Use after a change to shared clients, logging or retry code. tools: Read, Grep, Glob, Write, Bash
You write proofs, not production code. Only create or edit `*.test.ts` files.
Use the techniques in `labs/` (see `docs/writing-proofs.md`):
- Parity (
labs/01-environment-parity): run the same input through dev, staging and prod and ass
Alternatives in Testing & QA
- Helper Author — Owns the shared Cypress helper library for component automation 41k ★
- Claude Code Test Agent — Tests all 27 Claude Code hooks by logging each event to tests-agents-hook/agent-hook-fired.log 321 ★
- Staff Code Reviewer — Use for adversarial pre-merge code review 179 ★
Full documentation available on GitHub
View Source RepositoryRelated Agents
Agency QA
Acceptance verifier and red-proof sabotage tester for one stage (standard repos and Unity EditMode/PlayMode).
Test And Prove Agent
Evidence owner for redacted, reproducible browser proof bundles.
Post Merge QA
Post-merge QA on develop or develop- . Confirmed scope, environment preflight, flow/UX exercise, optional desi
Claude Skill Builder
Builds Claude Code skills in ~/.claude/skills/ / (SKILL.md, scripts, tests), incl. API helpers with secret han
Spgr Agent Documentation
Owns all developer-facing documentation: API reference, SDK clients, README, changelog, release notes, docstri
Qe Test Executor
Multi-framework test executor with parallel execution, retry logic, and real-time reporting
Related Skills
AI Agent Tool Calling
Python AI agent with Claude tool-calling, FastAPI endpoint, retry logic with exponential backoff, per-tool err
Crypto MCP Server By Corax CoLAB
An Autonomous Hedge Fund AI Orchestrator by Corax CoLAB. Features Multi-Agent Consensus (Gemini, Claude, OpenA
Agent Observability Logging
Log everything that happens inside your agent system, from deterministic code: hook-written audit events, inva