React Developer

STOP - Read before ANY tool call

1. Read E:/Source/Repos/agent-dev-toolkit/scripts/validation/fixtures/codex/rules/guardrails.md
2. Read _shared/sdd-contracts/SESSION.md; load session-state for $Cwd
3. If the relevant gate is not approved: STOP - ask user (pt-BR) - do NOT Write/Shell
4. SDD/develop skills: after ONE step/task, STOP session - handoff only
5. This skill body is English; user-facing prompts may be (pt-BR)

Step -1 - Gate check (report in chat before continuing)

Gate check:
[ ] guardrails.mdc read
[ ] SESSION.md read; session-state loaded
[ ] User confirmed current action (sim)
-> If any unchecked: STOP

Trigger

Invoke when the user asks for: /react-developer, react fix, or small isolated React (web) work that does not need a full PRD/PLAN cycle.

Outcome

Working React components and tests in the target workspace, validated with tests/build, on a valid feature branch, with optional /commit handoff. Does not replace SDD for multi-step or cross-repo features.

When to prefer SDD instead

Recommend /sdd-spec -> sdd-plan -> sdd-develop if two or more apply:

DESIGN-BRIEF acceptance

If docs/DESIGN-BRIEF.md or docs/design/DESIGN-BRIEF.md exists, treat it as the acceptance source. Map sections to React/TSX; do not reinterpret visual decisions. Implement one session scope from section 10 only.

If the task is net-new UI without a brief, recommend /impeccable shape in a new session before implementing.

Lazy-load (only when needed)

Do not preload unrelated guideline trees. Do not redesign react-guidelines/ in this skill — load existing pack rows only. For mobile React Native / Expo, route to /react-native-developer instead.

Process

Step -1b - Caveman Mode (Full cap)

1. Read E:/Source/Repos/agent-dev-toolkit/scripts/validation/fixtures/codex/sdd/preferences.json (create { "caveman_mode": false, "caveman_level": "full" } if missing).
2. If caveman_mode is false: continue without compression.
3. If true: load E:/Source/Repos/agent-dev-toolkit/scripts/validation/fixtures/codex/plugin/skills/_shared/caveman/CAVEMAN.md; apply Full participation cap + prefs caveman_level (Lite skills never escalate); show once: [Caveman] Modo ativo (respostas compactas, level={effective}). Digite caveman off para desativar.
4. Honor caveman on|off|status|lite|full|ultra (and stop caveman / normal mode) during the session.
5. Auto-Clarity + never-compress gates/drafts/paths per CAVEMAN.md.

Subagent-first (before implement)

Classify complexity → consult capability subagents → if medium/complex and native: spawn ≤2 children (scoped paths + receipt); trivial stays in-parent; if subagents=none or Task unavailable → fallback in-parent (never hard-fail). Load SPAWN.md + subagent-first.md; do not paste guidelines into child prompts.

0. Workspace

Confirm React web project (package.json with React; not RN-only). If blip-ds is present, load Blip guidelines from lazy-load table. Follow step-0-context.md. Summarize acceptance.

1. Guidelines (step 0.5)

Follow step-0.5-review-guidelines.md: load only required React/frontend guideline files for this task.

2. Branch (step 3)

Baseline from user or repo default. Create/checkout feature/<slug> or feat/<id> — never commit on main / master / develop.

3. Plan micro-steps

List 3-7 concrete tasks; checkpoint per context-management.mdc (>= 40% -> pause, offer /commit).

4. Implement

Functional components, hooks, clean React architecture. Match existing patterns. Apply react-guidelines/ while writing — do not paste full bodies into chat.

5. Tests

Jest / React Testing Library for changed behavior.

6. Build and test

npm test
npm run build

(or project-equivalent scripts)

7. Pre-commit (step 3.5) and handoff

Run step-3.5-precommit-validation.md when appropriate. Offer /commit — do not commit automatically. Before push/PR, run step-7-checklist.md.

Must not

• Paste guideline packs into child prompts (pass scoped paths + require receipt only)
• Spawn children for trivial work (keep in-parent)
• Hard-fail when capability subagents is none or Task is unavailable (use fallback in-parent)
• Redesign or rewrite the react-guidelines/ pack in-session
• Auto-commit or auto-PR
• Leave AI traces in code or identifiers
• Use obsolete corporate pipeline docs

Handoff