tdmcp-test-coverage
Orchestrate any tdmcp test-coverage work: run the coverage harness, inspect coverage gaps, plan focused Vitest/msw or bridge tests, delegate coverage writers, verify coverage deltas, and re-run gates.
SKILL.md.claude/skills/tdmcp-test-coverage/SKILL.md
GitHubtdmcp-test-coverage
Use this skill for coverage-improvement work in this repository. The goal is more tested behavior, not prettier percentages.
Phase 0 - Context
- Read
CLAUDE.md,package.json,vitest.config.ts, and the nearest tests. - Run
npm run coverage:harnessunless the user asked for a static-only pass. - Read
_workspace/coverage/latest.md. - Treat the actionable scope as
src/**/*.ts;src/knowledge/data/**is generated reference data and is intentionally excluded.
Phase 1 - Select gaps
Choose one to three independent gaps. Prefer seams where a test can lock a real contract:
- CLI/config: parsing, defaulting, env precedence, diagnostics, and exit behavior.
- Resource/knowledge loaders: URI matching, malformed input, ranking, and stable result shape.
- Tool implementations: Zod defaults, bridge payloads, warnings, and
isErrorpaths using msw. - Server/client boundaries: offline TD behavior, auth forwarding, timeout/retry, event streaming, and transport setup.
Skip generated data, docs-only files, and import-only tests.
Phase 2 - Build tests
For parallel work, spawn tdmcp-coverage-writer agents with one seam each. Give
each writer:
- target file(s) and nearest existing test to mirror;
- behavior to assert;
- expected failure mode or branch to cover;
- exact narrow command to run.
The lead is the single writer for shared config, package scripts, and harness docs. Do not let multiple agents edit the same test file.
Phase 3 - Verify
Run narrow tests first. Then run:
npm run typecheck
npm run build
./node_modules/.bin/biome check .
npm run coverage:harness
npm run validate:recipes
npm run test:bridge
If a gate fails, fix forward. Do not lower coverage thresholds, delete assertions, or exclude executable code to pass.
Phase 4 - Report
Write _workspace/coverage/wave-<date>.md with:
- starting and ending coverage summary;
- tests added and behaviors covered;
- commands run;
- PASS/FAIL/UNVERIFIED buckets;
- next suggested gaps from
_workspace/coverage/latest.md.
Test scenarios
- Normal flow: coverage harness identifies
src/cli/chat.ts; writer adds focused CLI behavior tests; lead runs coverage and gates; report includes delta. - Error flow: a new test exposes a production bug; report the failing behavior, patch the smallest safe code path, keep the regression test, and re-run gates.