mk:nyquist
Test-to-requirement coverage mapping — reads plan acceptance criteria and test files, produces a coverage gap report showing which requirements have no tests.
What This Skill Does
Analyzes plan acceptance criteria against existing test files to identify coverage gaps — requirements in the plan with no corresponding test. Named after the Nyquist sampling theorem: sufficient test coverage prevents aliased (missed) requirements.
This is a read-only analysis skill. It never modifies code or plans.
When to Use
- Phase 2 (Test): After tester writes initial tests — verify all acceptance criteria are covered before implementation.
- Phase 4 (Review): As part of test coverage dimension — verify no gaps before shipping.
- Before Gate 2: Verify implementation matches plan.
- Triggers: "check test coverage", "are all requirements tested", "coverage gaps", "nyquist".
NOT for: running tests (mk:testing), structural code review (mk:review), writing tests (tester agent's job).
Core Capabilities
5-Step Workflow
- Load plan — Read
tasks/plans/YYMMDD-name/plan.md, extract all acceptance criteria. - Scan tests — Glob for test files:
**/*.test.*,**/*.spec.*,**/test_*.*. - Map criteria to tests — For each acceptance criterion, search test strings for matching keywords.
- Identify gaps — Criteria with no matching test = coverage gap.
- Produce report — Structured output with mapping table and gap list.
Matching Strategy
| Strategy | How it works |
|---|---|
| Keyword matching | Extract key nouns/verbs from criteria, search in describe/it/test strings |
| File path matching | If criterion mentions "auth", search in **/auth*test* |
| Fuzzy matching | If exact match fails, check semantic equivalents (e.g., "login" ≈ "authenticate") |
| Manual override | If no match found, ask user to confirm gap is real (not a naming mismatch) |
Output Format
Example Prompt
/mk:nyquist Before we start building the user onboarding flow, check if all acceptance criteria from the plan have corresponding tests. Show me the coverage gap report — which requirements have no test at all and which have only partial coverage.
mk:elicit
Structured second-pass reasoning — re-examine review verdicts, plans, or analyses through named methods (pre-mortem, red team, Socratic, inversion, first principles, and more).
mk:cso
Chief Security Officer — infrastructure-first security audit. Two modes: daily (zero-noise, 8/10 confidence gate) and comprehensive (monthly deep scan, 2/10 bar). Covers secrets archaeology, dependency supply chain, CI/CD pipeline, LLM/AI security, OWASP Top 10, STRIDE threat modeling, and active verification.