tdd-guardian

Plugin tdd-guardian in xiaolai/codex-marketplace. Commands read from the repo's manifest — 1 month ago.

Skills (32)

• tdd-guardian-audit-coverage

  Run the tdd-guardian-coverage-auditor skill to run the coverage command, compare against thresholds, and list uncovered branches with proposed tests.

• tdd-guardian-audit-mutation

  Run the tdd-guardian-mutation-auditor skill to run mutation testing and list surviving mutants. Skips cleanly if `requireMutation=false` or the tool is not installed.

• tdd-guardian-coverage-auditor

  Enforce strict coverage gates and identify exact missing coverage scenarios. Use it to check the configured coverage thresholds, or to audit coverage for one file whose new branches may be untested; it runs the coverage lanes and proposes a test for each gap. Not for writing the missing tests (tdd-g

• tdd-guardian-coverage-gate

  Coverage thresholds plus test quality: coverage without behavioral assertions fails.

• tdd-guardian-design-tests

  Run the tdd-guardian-test-designer skill to produce a concrete behavior-driven test matrix for a plan. Rejects wiring-only designs. Does NOT write implementation code.

• tdd-guardian-gate

  Run the lanes bound to a trigger (commit, push, taskCompleted, manual) or a named lane, then evaluate the coverage and mutation gates and refresh gate state. This is how you make a stale gate fresh before committing or pushing.

• tdd-guardian-implement

  Run the tdd-guardian-implementer skill for ONE work item — red, green, refactor — then verify with the project's test command.

• tdd-guardian-implementer

  Implement planned work items in small batches with test-first discipline, one work item at a time (for example \"implement WI-2\"): failing tests first, then the minimal implementation, verified after each batch. Not for planning a feature (use tdd-guardian-planner), and not for implementing many wo

• tdd-guardian-init

  Detect every test lane in this repository — unit, integration, e2e, contract — verify each by dry-run probe, and write `.claude/tdd-guardian/config.json`. Reads CI config first, since that is what maintainers actually run.

• tdd-guardian-lane-policy

  Test lanes (unit, integration, e2e, contract): what goes where, triggers, coverage.

• tdd-guardian-mutation-auditor

  Validate test robustness using mutation testing and report every surviving mutant with the test that would kill it, including whether the mutation score meets the kill-rate threshold and whether the mutation tool is installed. Report-only — the implementer writes the tests. Not for checking coverage

• tdd-guardian-mutation-gate

  Mutation testing gate: per-language tools (Stryker, PIT, mutmut), weak assertions.

• tdd-guardian-plan

  Break a TDD task into work items, acceptance criteria, and required test targets, using the tdd-guardian planner, and save the plan under .claude/tdd-guardian/. Does NOT write code or tests.

• tdd-guardian-planner

  Break a request — a new feature or a refactor — into implementation work items with explicit acceptance criteria, risks, and test targets, before any code is written. Not for implementing code (use tdd-guardian-implementer) or for coverage questions (use tdd-guardian-coverage-auditor).

• tdd-guardian-policy-core

  Core TDD policy: plan-first development, assertion and spec levels, completion gates.

• tdd-guardian-probe

  Dry-run every configured lane to verify its command resolves and discovers tests, without executing the suites. Read-only; runs no tests and writes no files.

• tdd-guardian-review-gate

  Findings-first code review: severity order, test-gap findings, test-quality audit.

• tdd-guardian-review

  Run the tdd-guardian-reviewer skill for a final code + test-quality review. Classifies every expect() call as behavior vs wiring, flags mocked internal modules, and reports missing tests.

• tdd-guardian-reviewer

  Final reviewer that audits code quality, test quality (wiring vs behavior), and coverage gaps; also use it to flag the wiring-only tests in a specific test file. Report-only. Not for fixing the tests it flags (use tdd-guardian-implementer) or for running coverage (use tdd-guardian-coverage-auditor).

• tdd-guardian-spec-adversary

  Attack a test matrix before any implementation exists — find the wrong implementation that would pass every case, such as one that skips a state transition or breaks an invariant. Report-only; the designer fixes the matrix. Not for reviewing tests after the code is written (use tdd-guardian-reviewer

• tdd-guardian-status

  Read-only status report — shows last-run coverage %, mutation score, review verdict, and per-work-item state from `.claude/tdd-guardian/state.json`. Does NOT run any auditor skill, test, or gate.

• tdd-guardian-test-designer

  Design behavior-driven tests with explicit assertion strategies — success, boundary, invalid-input, edge and concurrency cases, using real fixtures rather than mocking internal modules. Rejects wiring-only test designs. Not for writing implementation code (use tdd-guardian-implementer) or for runnin

• tdd-guardian-test-matrix

  Test matrix for changed behavior, with an explicit assertion strategy per case.

• tdd-guardian-tooling-catalog

  Per-language test runners, coverage tools, mutation tools and dry-run probe commands.

• tdd-guardian-workflow

  Run strict TDD orchestration by chaining the six focused skills: plan → design-tests (with adversarial attack on the matrix) → implement (per WI, with red receipts) → audit-coverage → audit-mutation → review. Halts immediately on any gate failure. No commits before green.

• coverage-gate

  Coverage thresholds plus test quality: coverage without behavioral assertions fails.

• lane-policy

  Test lanes (unit, integration, e2e, contract): what goes where, triggers, coverage.

• mutation-gate

  Mutation testing gate: per-language tools (Stryker, PIT, mutmut), weak assertions.

• policy-core

  Core TDD policy: plan-first development, assertion and spec levels, completion gates.

• review-gate

  Findings-first code review: severity order, test-gap findings, test-quality audit.

• test-matrix

  Test matrix for changed behavior, with an explicit assertion strategy per case.

• tooling-catalog

  Per-language test runners, coverage tools, mutation tools and dry-run probe commands.

Slash commands (16)

• /tdd-guardian:audit-coverage

  Dispatch the tdd-coverage-auditor agent to run the coverage command, compare against thresholds, and list uncovered branches with proposed tests.

• /tdd-guardian:audit-mutation

  Dispatch the tdd-mutation-auditor agent to run mutation testing and list surviving mutants. Skips cleanly if `requireMutation=false` or the tool is not installed.

• /tdd-guardian:design-tests

  Dispatch the tdd-test-designer agent to produce a concrete behavior-driven test matrix for a plan. Rejects wiring-only designs. Does NOT write implementation code.

• /tdd-guardian:gate

  Run the lanes bound to a trigger (commit, push, taskCompleted, manual) or a named lane, then evaluate the coverage and mutation gates and refresh gate state. This is how you make a stale gate fresh before committing or pushing.

• /tdd-guardian:implement

  Dispatch the tdd-implementer agent for ONE work item — red, green, refactor — then verify with the project's test command.

• /tdd-guardian:init

  Detect every test lane in this repository — unit, integration, e2e, contract — verify each by dry-run probe, and write `.claude/tdd-guardian/config.json`. Reads CI config first, since that is what maintainers actually run.

• /tdd-guardian:plan

  Dispatch the tdd-planner agent to break a task into work items, acceptance criteria, and required test targets. Does NOT write code or tests.

• /tdd-guardian:probe

  Dry-run every configured lane to verify its command resolves and discovers tests, without executing the suites. Read-only; runs no tests and writes no files.

• /tdd-guardian:review

  Dispatch the tdd-reviewer agent for a final code + test-quality review. Classifies every expect() call as behavior vs wiring, flags mocked internal modules, and reports missing tests.

• /tdd-guardian:shared:detect-tooling

  Shared: detect test lanes, runners, coverage tools, and e2e suites from CI config, manifests, and test topology — then verify each by dry-run probe

• /tdd-guardian:shared:load-config

  Shared: load .claude/tdd-guardian/config.json, migrate v1 to v2, validate lanes, fail loudly if missing

• /tdd-guardian:shared:parse-coverage

  Shared: parse and merge coverage across 9 formats — Istanbul, coverage.py, LCOV, Cobertura, JaCoCo, Clover, go-cover, SimpleCov

• /tdd-guardian:shared:parse-mutation

  Shared: parse mutation testing output into score + survivor list across the Stryker family, PIT, Infection, mutmut, go-mutesting, cargo-mutants, and any tool emitting the mutation-testing-elements schema

• /tdd-guardian:shared:run-lane

  Shared: execute one lane (setup, command, coverage report, teardown) and classify the outcome, distinguishing runner failure from test failure

• /tdd-guardian:status

  Read-only status report — shows last-run coverage %, mutation score, review verdict, and per-work-item state from `.claude/tdd-guardian/state.json`. Does NOT run any agent, test, or gate.

• /tdd-guardian:workflow

  Run strict TDD orchestration by chaining the six focused commands: plan → design-tests (with adversarial attack on the matrix) → implement (per WI, with red receipts) → audit-coverage → audit-mutation → review. Halts immediately on any gate failure. No commits before green.

Subagents (7)

• tdd-coverage-auditor

  Enforce strict coverage gates and identify exact missing coverage scenarios. Use it to check the configured coverage thresholds, or to audit coverage for one file whose new branches may be untested; it runs the coverage lanes and proposes a test for each gap. Not for writing the missing tests (tdd-i

• tdd-implementer

  Implement planned work items in small batches with test-first discipline, one work item at a time (for example "implement WI-2"): failing tests first, then the minimal implementation, verified after each batch. Not for planning a feature (use tdd-planner), and not for implementing many work items in

• tdd-mutation-auditor

  Validate test robustness using mutation testing and report every surviving mutant with the test that would kill it, including whether the mutation score meets the kill-rate threshold and whether the mutation tool is installed. Report-only — the implementer writes the tests. Not for checking coverage

• tdd-planner

  Break a request — a new feature or a refactor — into implementation work items with explicit acceptance criteria, risks, and test targets, before any code is written. Not for implementing code (use tdd-implementer) or for coverage questions (use tdd-coverage-auditor). <example> Context: User asks to

• tdd-reviewer

  Final reviewer that audits code quality, test quality (wiring vs behavior), and coverage gaps; also use it to flag the wiring-only tests in a specific test file. Report-only. Not for fixing the tests it flags (use tdd-implementer) or for running coverage (use tdd-coverage-auditor). <example> Context

• tdd-spec-adversary

  Attack a test matrix before any implementation exists — find the wrong implementation that would pass every case, such as one that skips a state transition or breaks an invariant. Report-only; the designer fixes the matrix. Not for reviewing tests after the code is written (use tdd-reviewer) or for

• tdd-test-designer

  Design behavior-driven tests with explicit assertion strategies — success, boundary, invalid-input, edge and concurrency cases, using real fixtures rather than mocking internal modules. Rejects wiring-only test designs. Not for writing implementation code (use tdd-implementer) or for running coverag

Hooks (2)

• PreToolUse

  node "${CLAUDE_PLUGIN_ROOT}/scripts/tdd-guardian/pretool_guard.js"

• TaskCompleted

  node "${CLAUDE_PLUGIN_ROOT}/scripts/tdd-guardian/taskcompleted_gate.js"

Similar plugins