codeGrade S · 93/100 Safety 100

Test Forger

Tests that actually catch regressions

Generates boundary-focused tests from implementation code: happy path, edges, failure modes — plus the one test the author hoped nobody would write.

unit test generationedge case designmutation review

Works with: OpenClaw · Claude Code · Cursor · Any MCP agent · v1.6.1 · MIT · by skillforge-curated

SKILL.md preview

# Test Forger

## Triggers
User asks for tests for a file or function.

## Workflow
1. Read the implementation; list its branches and boundaries.
2. Write happy-path, edge, and failure-mode tests.
3. Add one adversarial test per public function.
4. Run mentally; fix flakiness before emitting.

## Guardrails
- Tests must be deterministic — no sleeps, no network.
- Match the repo test framework.

Install in one command

1. Create the skill folder: mkdir -p skills/test-forger

2. Install: curl -fsSL "https://skillforge-aniruddha-adaks-projects.vercel.app/api/skills/test-forger/skill.md" -o skills/test-forger/SKILL.md

3. Mention it in SOUL.md or BOOTSTRAP.md so OpenClaw loads it on boot.

Install command

curl -fsSL "https://skillforge-aniruddha-adaks-projects.vercel.app/api/skills/test-forger/skill.md" -o skills/test-forger/SKILL.md

Why this score

Documentation16/20

3/3 required SKILL.md sections, 397 chars

Capability breadth12/15

3 declared capabilities

Agent compatibility15/15

Works with openclaw, claude-code, cursor, generic-mcp

License openness10/10

MIT — reuse freely

Community signal20/20

15800 registry installs

Safety posture20/20

No risky capabilities declared

Audit seal

SHA-384(prev ‖ canonicalJson)

3d4a6a3179a74f0c508e104698fa9dae03f32c74506892d0036a5df832fe8b548ecb9742a1759ba617d99305720c2f47

chain head 99bddff1717e676698441b4a1963172e…

Replay: POST {ids} to /api/seals/verify

Pairs well with

PR Reviewer95/100 SQL Smith95/100 Regex Wizard93/100