meisijiya/skills

1

debugging-and-error-recovery Use when tests fail, builds break, or behavior is unexpected. Runs a five-step triage protocol to localize the failur…

meisijiya/skills

29

2

source-driven-development Use when the API is unfamiliar, version-sensitive (major upgrade), or behavior is unexpectedly wrong. Verifies framew…

meisijiya/skills

29

3

api-and-interface-design Use when designing REST/GraphQL/RPC endpoints, defining module type contracts, establishing frontend↔backend boundari…

meisijiya/skills

28

4

build-gate-visual-review Use when user requests Markdown/text design-alignment review, or a responsive HTML page (teacher-skill overlay) for l…

meisijiya/skills

28

5

designer-handoff Use when project has UI and frontend agent must follow design rules (color, typography, anti-patterns) instead of gue…

meisijiya/skills

28

6

documentation-and-adrs Use when decision is irreversible + cross-time + cross-team (data model / API contract / dependency upgrade / depreca…

meisijiya/skills

28

7

incremental-implementation Use when a change touches more than one file, has 3+ steps, or involves refactoring/migrating existing code. Decompos…

meisijiya/skills

28

8

observability-and-instrumentation Use when shipping any production feature, telemetry missing for diagnosis, or adding telemetry to new service. Instru…

meisijiya/skills

28

9

performance-optimization Use when production p99 or p95 regressed, k6 gate fired, or user-perceived latency high. Optimizes backend, database,…

meisijiya/skills

28

10

security-and-hardening Use when writing or reviewing code handling user input, credentials, PII, auth, storage, third-party integrations. Ha…

meisijiya/skills

28

11

spec-driven-development Use when starting a new project, new feature, or significant change; refactoring across modules; or any work touching…

meisijiya/skills

28

12

test-driven-development Use when implementing new logic, fixing any bug, or changing behavior. Enforces red-green-refactor: write a failing t…

meisijiya/skills

28

13

using-meisijiya-skills Use when starting any session or about to take action in a project with meisijiya-skills installed. Routes work via d…

meisijiya/skills

28

14

verification-before-completion Use when about to claim work is complete, fixed, or passing, before committing, creating PRs, or telling the user 'do…

meisijiya/skills

28

15

brainstorming Use when starting a non-trivial feature, behavior change, or when a proposed plan's intent is unclear. Presents desig…

meisijiya/skills

27

16

writing-skills Use when creating a new skill, editing an existing skill, or extracting a repeated/abstractable workflow into a reusa…

meisijiya/skills

27

17

improve-codebase-architecture Use when onboarding to unfamiliar code, doing a weekly architecture review, or agent output quality has degraded. Sur…

meisijiya/skills

25

18

verify-chain Role-adversarial tech article verification: Critic extracts claims, parallel Verifier subagents web-check, Repairer a…

meisijiya/skills

25

19

ai-code-blindspots Use when AI-generated or AI-modified code, post-Edit verification, or before completion claims. Reviews AI code for m…

meisijiya/skills

23

20

loop-me Use when user invokes /loop-me, wants a repeated workflow extracted into a spec, or formalizes a multi-step activity.…

meisijiya/skills

23

21

security-devsecops Use when adding deps, rotating secrets, configuring CI/CD, writing IaC, preparing production deploys. Secures supply …

meisijiya/skills

23

22

security-incident-response Use when actual or suspected breach is found via alert, user report, CVE disclosure, leaked secrets. Coordinates post…

meisijiya/skills

23

23

contract-strengthening Use after Spec attestation when change touches shared/cached state, time/clocks, concurrency, external boundaries, or…

meisijiya/skills

22

24

slice-review Use when a vertical slice in incremental-implementation finishes, per-slice spec-compliance and code-quality review. …

meisijiya/skills

22

25

closed-loop-delivery Use when change feels done and impulse is merge, users report bugs despite green CI, or auditing Definition of Done. …

meisijiya/skills

21

26

diagnosing-bugs Use when 5-step triage stalls to diagnose the root cause of a hard bug or performance regression. Forms ≥3 hypotheses…

meisijiya/skills

20

27

gha-security-review Use when adding, modifying, or auditing GitHub Actions workflows under .github/workflows/. Audits permissions, expres…

meisijiya/skills

20

28

k6-load-testing Use when before merging perf-sensitive changes, setting latency or error budgets, or proving a release holds load. Ru…

meisijiya/skills

20

29

pre-ship-gate Use when before any production rollout, after exit 0 deploy with silent failure, or canary stuck at 0%. Runs read-onl…

meisijiya/skills

20

30

production-incident-playbook Use when responding to active production incident, writing blameless postmortem, or coordinating in-flight mitigation…

meisijiya/skills

20

31

security-ownership-map Use before risky refactors, after incidents, when hiring for coverage gaps, or designing rotation/on-call. Maps peopl…

meisijiya/skills

20

32

security-threat-model Use when designing cross-boundary features, integrating third parties, refactoring auth, or expanding blast radius. T…

meisijiya/skills

20

33

stack-security-coder Use when writing or reviewing frontend, backend, or mobile code, or when audits reveal stack-layer gaps. Enforces per…

meisijiya/skills

20

34

supply-chain-risk-auditor Use when adding deps, during dep review, before lockfile, or when maintenance signals degrade. Audits trustworthyness…

meisijiya/skills

20

35

test-guard Use when reviewing AI-generated tests before merge, silently-green suite lets bugs escape, or auditing test quality. …

meisijiya/skills

20

36

teacher-skill Use when user says @teacher or Teaching-deck, asks for a learning doc or teaching deck, or needs pedagogy overlay on …

meisijiya/skills

18

37

meisijiya-minimalist-ui Use when a brief names a Linear, Notion, editorial, utilitarian, or minimalist aesthetic. Enforces premium minimalist…

meisijiya/skills

17

38

meisijiya-frontend-taste Use when writing marketing-grade landing pages or portfolio UI (React/Vue/Svelte/Tailwind), greenfield frontend. Enfo…

meisijiya/skills

16

39

meisijiya-redesign-ui Use when upgrading existing web/mobile UI to premium quality, code that already ships. Audits existing UI for AI-defa…

meisijiya/skills

16

40

prototype Use when Phase 1.2 spec has `[PROTO-RESOLVE]` marker that text alone cannot resolve (layout, density, motion). Genera…

meisijiya/skills

13

41

research Use when planning/design question needs cited evidence from official docs, RFCs, source repos, or project ADRs (requi…

meisijiya/skills

13

42

wayfinder Use when project scope exceeds one brainstorming session, or user wants decisions tracked across multiple sessions. O…

meisijiya/skills

13

43

meisijiya-phase-checkpoint Use when the agent detects a phase boundary in `.omo/plans/<slug>.md` — soft-prompt user with 'Phase N 完成,要 /hando…

meisijiya/skills

6

44

pwf-enforcer Hard-enforces planning-with-files (PWF) workflow on OpenCode via a TypeScript plugin (hard layer) plus a concise AGEN…

meisijiya/skills

6

45

code-simplification Backward-compat alias. For cleanup passes, prefer OMO built-in `refactor` (auto-routes to /simplify), `ponytail-revie…

meisijiya/skills

5

46

interview-me Backward-compat alias. The one-question-at-a-time discipline lives in `brainstorming` (core/). For new sessions, invo…

meisijiya/skills

5

47

meisijiya-handoff Use when user invokes /handoff, current session ends, or context must carry into a new session. Writes the cross-sess…

meisijiya/skills

4

48

andrej-karpathy-perspective Karpathy 思维框架(5 心智模型 + 8 启发式)分析 AI/学习/工程/教育/agent loop/Software 3.0。触发:Karpathy 视角、from-scra…

meisijiya/skills

1