Electron 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-artifacts/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

Use when user asks for /electron-developer, electron fix, or a small isolated Electron implementation.

Outcome

Working main/preload/renderer changes, validated with build and documented smoke (app launch), with optional handoff to /commit.

Renderer stack (orchestration)

Detect renderer framework from package.json:

Stay in electron-developer identity - load stack guidelines for UI patterns only; do not switch to react-developer / vue-developer.

When to escalate to SDD

Recommend sdd-spec -> sdd-plan -> sdd-develop if two or more apply: main+renderer+packaging overhaul, auto-update pipeline, cross-repo impact, 10+ files, or existing approved PLAN.

DESIGN-BRIEF acceptance

If docs/DESIGN-BRIEF.md or docs/design/DESIGN-BRIEF.md exists, treat it as the acceptance source. Map sections to renderer UI; 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 references

Do not preload unrelated guideline trees. Load only the electron-guidelines rows needed for the current task.

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 Electron project (electron, electron-builder, or electron-vite in package.json). Identify main/preload/renderer entry points.

1. Guidelines

Load Electron + renderer guidelines for this task. Review security defaults before IPC changes.

2. Branch

Use feature/<slug> or feat/<id>. Never commit on main/master/develop.

3. Micro-plan

Define 3-7 concrete tasks; checkpoint context at >= 40%.

4. Implement

Main/preload/renderer separation, typed IPC, contextIsolation. Match existing electron-vite or electron-builder layout.

5. Tests

Unit tests for pure modules; manual smoke: app launches, changed flow works.

6. Validate

npm run build

Document smoke steps in chat (launch app, exercise changed feature).

7. Handoff

Offer /commit. Do not commit automatically.

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)
• Enable nodeIntegration: true in renderer without explicit user approval
• Expose raw ipcRenderer on window without contextBridge
• Auto-commit or auto-PR
• Leave AI traces in code or identifiers

Handoff