Skip to content
Back to search

Claude-skills-v4

SkillWorks with Claude CodeGitHub

Claude Skills v4 — AI Engineering Hackathon

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today
Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/akramrafid--claude-skills-v4/badge.svg)](/listings/akramrafid--claude-skills-v4)

HTML

<a href="/listings/akramrafid--claude-skills-v4"><img src="/api/listings/akramrafid--claude-skills-v4/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

README.md · README · 53 lines
  1. # Claude Skills v4 — AI Engineering Hackathon
  2. 22 skills for Claude Code (also read by OpenCode and Antigravity): one for each of the 16 hackathon roles,
  3. plus 6 that run the team and review the code to production grade. Each role skill names its hand-off files
  4. and a runnable "done" check (`node .orca-team/tools/gates.mjs <check>`).
  5. | Hat | Skills |
  6. |---|---|
  7. | Frame | `product-requirements`, `ux-flow-design` |
  8. | Design | `solution-architecture`, `system-design`, `ai-engineering`, `data-engineering`, `database-engineering` |
  9. | Build | `tech-lead-integration`, `backend-engineering`, `frontend-engineering` |
  10. | Prove | `eval-engineering`, `sqa-testing`, `ai-red-team` |
  11. | Ship | `vercel-deploy`, `llmops-observability`, `demo-docs` |
  12. | Review & run | `code-review`, `security-review`, `production-hardening`, `production-standards`, `swarm-lead`, `credentials-gate` |
  13. - **[INSTALL.md](INSTALL.md)** — copy-paste PowerShell steps (project or all projects).
  14. - **[SKILLS-INDEX.md](SKILLS-INDEX.md)** — every role → skill → hand-off → check.
  15. - `skills/` — the skill folders (`SKILL.md` + `references/` code templates).
  16. - `orca-team-tools/` — the check runner (`gates.mjs`), test-server helper, Vercel env tool, env checker.
  17. Quick install into a project:
  18. ```powershell
  19. git clone https://github.com/akramrafid/Claude-skills-v4 "$env:TEMP\claude-skills-v4"
  20. Copy-Item "$env:TEMP\claude-skills-v4\skills\*" ".claude\skills\" -Recurse -Force
  21. Copy-Item "$env:TEMP\claude-skills-v4\skills\*" ".agents\skills\" -Recurse -Force
  22. ```
  23. Run those from the project folder; see INSTALL.md for the check tools and for removing old v3 skills.
  24. ## Which model plays which seat
  25. Skills name **seats**, not models. One file maps seats to models:
  26. [`skills/swarm-lead/references/model-routing.md`](skills/swarm-lead/references/model-routing.md) — change models there only.
  27. | Seat | Work | Model (effort) | Out of quota → |
  28. |---|---|---|---|
  29. | LEAD | requirements, architecture, integration | Claude Opus 5.5 (max) | Gemini 4 Neon max → Muse Wonder max |
  30. | DESIGN | system design, PRD/architecture critique | Gemini 4 Neon (max) | Muse Wonder max → Opus max |
  31. | UI | UX, frontend | Gemini 4 Neon (xhigh) · second UI slice: Gemini 4 Hydrogen (xhigh) | Hydrogen xhigh → Muse Wonder max |
  32. | CORE | backend, AI engineering, LLMOps | Claude Sonnet 5.5 (max) | Gemini 4 Neon xhigh → Kimi K3.1 Pro max |
  33. | DATA | data + database | Kimi K3.1 Pro (max) | Gemini 4 Neon xhigh → Muse Wonder max |
  34. | QA | evals, tests, deploy prep, docs | Kimi K3.1 Pro (max) | Gemini 4 Neon xhigh → Hydrogen xhigh |
  35. | REVIEWER-CODE | code review | Claude Opus 5.5 (max) | Gemini 4 Neon max |
  36. | REVIEWER-SEC | security review, red team | Muse Wonder (max) — never Kimi | Gemini 4 Neon max |
  37. | OUTSIDER | hallway test, dry run | Muse Wonder (max) | Gemini 4 Hydrogen xhigh |
  38. **Any model out of quota → Gemini 4 Neon** (xhigh to build, max to design and review). A checker never shares
  39. the author's model family (Anthropic · Google · Meta · Moonshot).
  40. ## Architecture the skills build
  41. Modular monolith · **vertical slices** (`apps/api/app/features/<name>/`, `apps/web/src/features/<name>/`) ·
  42. **ports & adapters** around the LLM, embeddings, vector store and database. One MUST feature = one slice = one
  43. backend worker + one frontend worker, so parallel agents never edit the same files. `gates.mjs slices` (also run
  44. inside `gates.mjs merge`) fails when a feature imports another feature or a service imports FastAPI, an SDK or
  45. the ORM. Full rules: `skills/solution-architecture/references/blueprint.md` §3.

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.