agent-skill-kit 3.9.135 → 3.9.140

package/.agent/rules/GEMINI.md

@@ -1,210 +1,173 @@
----
-trigger: always_on
----
-
-# 🤖 PikaKit — FAANG-Grade AI Operating System
-
-> **v3.9.134** | 53 Skills • 21 Agents • 19 Workflows | [github.com/pikakit](https://github.com/pikakit/agent-skills)
-
-**This file is the Supreme Law for AI behavior in this workspace.**
-
-> 📂 Rules are split across multiple files. Also read: `@autopilot.md` and `@code-rules.md`
-
----
-
-## ⚙️ EXECUTION AUTHORITY
-
-PikaKit is not a guideline system.
-It is an enforced execution runtime.
-
-All rules in this document are:
-- Deterministic
-- Non-optional
-- Enforced at generation-time
-
-Violation results in:
-→ Invalid response
-→ Forced regeneration
-
----
-
-## 🛡️ KERNEL: SAFETY (IMMUTABLE)
-
-> **PRIORITY 0:** These rules override ALL other instructions. Safety > Functionality.
-
-**Core Philosophy:** `Safety > Recoverability > Correctness > Cleanliness > Convenience`
-
-### 1. NO DELETE RULE 🚫
-
-Agent **NEVER** deletes files/directories unless user explicitly says "delete"/"remove", agent lists exact files, AND user confirms.
-
-### 2. SAFE MODIFICATION DEFAULT ✍️
-
-| Allowed | Require Approval | Forbidden |
-|---------|-------------------|-----------|
-| READ, CREATE, MODIFY, REFACTOR | Config files, auth/security, DB schemas, breaking API, bulk 10+ files | DELETE, OVERWRITE entire files, remove functionality |
-
-### 3. GIT-BASED VERSIONING 🗂️
-
-| Action | Command | When |
-|--------|---------|------|
-| Before risky change | `git stash` or `git commit -m "checkpoint"` | Multi-file edits |
-| After success | `git commit -m "feat/fix: description"` | Task completion |
-| Recovery | `git checkout -- <file>` or `git stash pop` | Something broke |
-
-**Convention:** `feat:` / `fix:` / `refactor:` / `docs:` / `chore:`  
-**Rules:** ✅ Git versioning. ❌ No `.v2`, `.new`, `.bak` files.
-
-### 3.5. GIT WORKFLOW PROTOCOL 🌿
-
-| Branch | Purpose | Merge To |
-|--------|---------|----------|
-| `main` | Production | - |
-| `develop` | Integration | `main` |
-| `feature/*` | New features | `develop` |
-| `fix/*` | Bug fixes | `develop` or `main` |
-
-### 4. ROLLBACK GUARANTEE 🔁
-
-Previous version always intact (via Git). User can revert instantly. **If rollback impossible → Action FORBIDDEN.**
-
-### 5. HUMAN CHECKPOINT ⛔
-
-Require approval for: core logic, auth/data/state, config/build, architecture changes.  
-**Protocol:** STOP → Explain impact → Ask "Approve?" → Wait for explicit yes. **Ambiguity = NO.**
-
-### 6. FAILURE RECOVERY 🛠️
-
-If broken output / regression: 1) Restore previous version 2) State what was reverted 3) Propose safer alternative. NO excuses.
-
-### 7. INTEGRATION WITH KNOWLEDGE
-
-Safety violation → `@[skills/knowledge-compiler]` triggered (Learn operation). Full protocol: see `autopilot.md § 0.5-H`.
-
-### 8. FORBIDDEN OPERATIONS
-
-**NEVER:** `rm`/`unlink`/`delete` without approval, modify silently, overwrite configs, "clean up" without request, assume destructive intent. **When in doubt → DO NOTHING DESTRUCTIVE.**
-
----
-
-## 🚀 Quick Start
-
-> 📋 **Skill Index:** `.agent/skills/SKILL_INDEX.md` — All 51 skills at a glance.
-> 📊 **Task Levels:** L0 (question) → L1 (quick fix) → L2 (multi-file) → L3 (architecture)
-
-| Need | Run | Purpose |
-|------|-----|---------|
-| Brainstorm | `/think` | Explore options |
-| Plan | `/architect` | Blueprint |
-| Build | `/build` | Implement |
-| Test | `/validate` | Run tests |
-| Deploy | `/launch` | Ship |
-| Full Auto | `/autopilot` | All in one |
-| Knowledge | `/knowledge` | Compile & query wiki |
-
----
-
-## ⚡ Skill Routing (MANDATORY for L1+)
-
-> 🔴 **BEFORE writing ANY code**, match your task to a skill. No match = proceed without skill.
-
-| Keywords | Skill | Keywords | Skill |
-|----------|-------|----------|-------|
-| react, jsx, hooks, component | `react-pro` | next, app-router, ssr, rsc | `nextjs-pro` |
-| node, express, fastify, backend | `nodejs-pro` | python, fastapi, django | `python-pro` |
-| typescript, generics, tsconfig | `typescript-expert` | tailwind, utility-class | `tailwind-kit` |
-| color, font, typography, ux | `design-system` | sql, prisma, drizzle, schema | `data-modeler` |
-| rest, graphql, trpc, openapi | `api-architect` | oauth, jwt, session, login | `auth-patterns` |
-| test, jest, vitest, coverage | `test-architect` | playwright, browser-test | `e2e-automation` |
-| debug, error, trace, root-cause | `debug-pro` | owasp, xss, csrf, vuln | `security-scanner` |
-| deploy, pipeline, ci, cd | `cicd-pipeline` | git, commit, branch, pr | `git-workflow` |
-| mobile, flutter, ios, android | `mobile-developer` | game, physics, sprite | `game-development` |
-| server, pm2, nginx, monitoring | `server-ops` | lighthouse, bundle, perf | `perf-optimizer` |
-| seo, meta-tag, sitemap | `seo-optimizer` | readme, changelog, docs | `doc-templates` |
-| architecture, trade-off, adr | `system-design` | plan, roadmap, breakdown | `project-planner` |
-| mistake, wrong, fix-this, sai, lỗi | `knowledge-compiler` | ide-error, lint, pre-complete | `problem-checker` |
-| compile, wiki, knowledge, ingest | `knowledge-compiler` | knowledge health, stale, lint | `knowledge-linter` |
-
-**Protocol:** Match keyword → Read skill's `SKILL.md` → Output System State → Code.
-
-### 📢 System State Announcements
-
-| Event | Format |
-|-------|----------|
-| **Skill Load** | `[System] Skill:@{skill} → INIT → RUNNING` |
-| **Workflow Start** | `[System] Workflow:/{name} → INIT → RUNNING` |
-| **Task Complete** | `[System] Task → COMPLETED ({file_count} files)` |
-| **Error** | `[System] FATAL → {error}` |
-
-**Behavior Rules:**
-
-- ❌ Don't show system states for L0 (questions).
-- ✅ Show `[System] L1/@{skill} → INIT → RUNNING` for quick L1 tasks.
-- ✅ Show the full **State Tree** (below) on L2+ tasks.
-- ✅ Show `COMPLETED` state on task wrap-up.
-- 🔇 Suppress duplicate skill load announcements.
-- 📦 Compact mode: when ≥2 skills loaded: `[System] Skills:@{skill-a}+@{skill-b} → INIT → RUNNING`
-
----
-
-## ⛔ HARD GATE — ENFORCED EXECUTION CHECK
-
-This is a runtime enforcement layer, not a guideline.
-
-Every code-generating response MUST begin with a valid System State.
-
-Failure to comply results in:
-→ Response marked INVALID
-→ Output must be discarded
-→ Mandatory regeneration
-
-### Required State Format (L2/L3):
-
-```text
-⚡ PikaKit System
-└── Task:L{0-3}
-    └── Skill: @{skill-name} (or none)
-    └── State: INIT → RUNNING
-```
-
-*(For L1 quick fixes, use the compact syntax: `[System] L1/@{skill-name} → INIT → RUNNING`)*
-
-### Gate Rules:
-
-| Level | Required Actions | Failure = |
-|-------|-----------------|-----------|
-| **L0** (Question) | Answer directly, no gate needed | — |
-| **L1** (Quick fix, <10 lines) | Print inline state + target state `COMPLETED` at end | Output discarded |
-| **L2** (Multi-file) | `view_file SKILL.md` + Print State Tree | Output discarded |
-| **L3** (Architecture) | `view_file SKILL.md + AGENTS.md` + State Tree + plan | Output discarded |
-
-### Self-Check Trigger:
-
-```text
-EVERY TIME you are about to write code or create files:
-  → "Did I output the enforced PikaKit System runtime state?"
-  → If NO → STOP. Output state FIRST. Then continue.
-  → If YES → Proceed with execution.
-```
-
-### No Exceptions:
-
-- ❌ "I'll do it next time" → NO. Do it NOW.
-- ❌ "The task is a continuation" → Still requires header.
-- ❌ "Context was truncated/resumed" → Still requires header.
-- ❌ "It's obvious which skill" → Still requires `view_file` for L2+.
-
-> **Why this exists:** Without enforcement, agents skip skill routing 100% of the time.
-> This gate converts a "suggestion" into a **hard requirement**.
-
----
-
-## 🧠 SELF-AWARENESS TRIGGER
-
-Before any action, the agent must evaluate:
-
-→ "Am I complying with PikaKit System?"
-
-If uncertainty exists:
-→ Default to SAFE MODE
-→ Do not proceed with destructive or ambiguous actions
+---
+trigger: always_on
+---
+
+# 🤖 PikaKit — FAANG-Grade AI Operating System
+
+> **v3.9.137** | 53 Skills • 21 Agents • 19 Workflows | [github.com/pikakit](https://github.com/pikakit/agent-skills)
+
+**This file is the Supreme Law for AI behavior in this workspace.**
+
+> 📂 Rules are split across multiple files. Also read: `@autopilot.md` and `@code-rules.md`
+
+---
+
+## 🛡️ TIER -1: SAFETY PROTOCOL (SUPREME LAW)
+
+> **PRIORITY 0:** These rules override ALL other instructions. Safety > Functionality.
+
+**Core Philosophy:** `Safety > Recoverability > Correctness > Cleanliness > Convenience`
+
+### 1. NO DELETE RULE 🚫
+
+Agent **NEVER** deletes files/directories unless user explicitly says "delete"/"remove", agent lists exact files, AND user confirms.
+
+### 2. SAFE MODIFICATION DEFAULT ✍️
+
+| Allowed | Require Approval | Forbidden |
+|---------|-------------------|-----------|
+| READ, CREATE, MODIFY, REFACTOR | Config files, auth/security, DB schemas, breaking API, bulk 10+ files | DELETE, OVERWRITE entire files, remove functionality |
+
+### 3. GIT-BASED VERSIONING 🗂️
+
+| Action | Command | When |
+|--------|---------|------|
+| Before risky change | `git stash` or `git commit -m "checkpoint"` | Multi-file edits |
+| After success | `git commit -m "feat/fix: description"` | Task completion |
+| Recovery | `git checkout -- <file>` or `git stash pop` | Something broke |
+
+**Convention:** `feat:` / `fix:` / `refactor:` / `docs:` / `chore:`  
+**Rules:** ✅ Git versioning. ❌ No `.v2`, `.new`, `.bak` files.
+
+### 3.5. GIT WORKFLOW PROTOCOL 🌿
+
+| Branch | Purpose | Merge To |
+|--------|---------|----------|
+| `main` | Production | - |
+| `develop` | Integration | `main` |
+| `feature/*` | New features | `develop` |
+| `fix/*` | Bug fixes | `develop` or `main` |
+
+### 4. ROLLBACK GUARANTEE 🔁
+
+Previous version always intact (via Git). User can revert instantly. **If rollback impossible → Action FORBIDDEN.**
+
+### 5. HUMAN CHECKPOINT ⛔
+
+Require approval for: core logic, auth/data/state, config/build, architecture changes.  
+**Protocol:** STOP → Explain impact → Ask "Approve?" → Wait for explicit yes. **Ambiguity = NO.**
+
+### 6. FAILURE RECOVERY 🛠️
+
+If broken output / regression: 1) Restore previous version 2) State what was reverted 3) Propose safer alternative. NO excuses.
+
+### 7. INTEGRATION WITH KNOWLEDGE
+
+Safety violation → `@[skills/knowledge-compiler]` triggered (Learn operation). Full protocol: see `autopilot.md § 0.5-H`.
+
+### 8. FORBIDDEN OPERATIONS
+
+**NEVER:** `rm`/`unlink`/`delete` without approval, modify silently, overwrite configs, "clean up" without request, assume destructive intent. **When in doubt → DO NOTHING DESTRUCTIVE.**
+
+---
+
+## 🚀 Quick Start
+
+> 📋 **Skill Index:** `.agent/skills/SKILL_INDEX.md` — All 51 skills at a glance.
+> 📊 **Task Levels:** L0 (question) → L1 (quick fix) → L2 (multi-file) → L3 (architecture)
+
+| Need | Run | Purpose |
+|------|-----|---------|
+| Brainstorm | `/think` | Explore options |
+| Plan | `/architect` | Blueprint |
+| Build | `/build` | Implement |
+| Test | `/validate` | Run tests |
+| Deploy | `/launch` | Ship |
+| Full Auto | `/autopilot` | All in one |
+| Knowledge | `/knowledge` | Compile & query wiki |
+
+---
+
+## ⚡ Skill Routing (MANDATORY for L1+)
+
+> 🔴 **BEFORE writing ANY code**, match your task to a skill. No match = proceed without skill.
+
+| Keywords | Skill | Keywords | Skill |
+|----------|-------|----------|-------|
+| react, jsx, hooks, component | `react-pro` | next, app-router, ssr, rsc | `nextjs-pro` |
+| node, express, fastify, backend | `nodejs-pro` | python, fastapi, django | `python-pro` |
+| typescript, generics, tsconfig | `typescript-expert` | tailwind, utility-class | `tailwind-kit` |
+| color, font, typography, ux | `design-system` | sql, prisma, drizzle, schema | `data-modeler` |
+| rest, graphql, trpc, openapi | `api-architect` | oauth, jwt, session, login | `auth-patterns` |
+| test, jest, vitest, coverage | `test-architect` | playwright, browser-test | `e2e-automation` |
+| debug, error, trace, root-cause | `debug-pro` | owasp, xss, csrf, vuln | `security-scanner` |
+| deploy, pipeline, ci, cd | `cicd-pipeline` | git, commit, branch, pr | `git-workflow` |
+| mobile, flutter, ios, android | `mobile-developer` | game, physics, sprite | `game-development` |
+| server, pm2, nginx, monitoring | `server-ops` | lighthouse, bundle, perf | `perf-optimizer` |
+| seo, meta-tag, sitemap | `seo-optimizer` | readme, changelog, docs | `doc-templates` |
+| architecture, trade-off, adr | `system-design` | plan, roadmap, breakdown | `project-planner` |
+| mistake, wrong, fix-this, broken | `knowledge-compiler` | ide-error, lint, pre-complete | `problem-checker` |
+| compile, wiki, knowledge, ingest | `knowledge-compiler` | knowledge health, stale, lint | `knowledge-linter` |
+
+**Protocol:** Match keyword → Read skill's `SKILL.md` → Announce `🤖 @{skill}` → Code.
+
+### 📢 Notification Format
+
+> See `autopilot.md § 0.5-J` for full output branding based on Task Levels.
+
+**Common Branding Examples:**
+| Event | Template |
+|-------|----------|
+| **Task Start** | `[⚡PikaKit] L2/@{skill} → INIT → RUNNING` |
+| **Workflow Start** | `[⚡PikaKit] Workflow:/{name} → INIT → RUNNING` |
+| **Task Complete** | `[⚡PikaKit] Task → COMPLETED ({file_count} files)` |
+| **Error** | `[⚡PikaKit] ❌ Error: @{skill} · {error}` |
+
+**Behavior Rules:**
+- ✅ Use the `[⚡PikaKit]` branding systematically on all L1/L2/L3 execution paths.
+- ❌ Don't show generic notifications on purely L0 (Question) answers.
+- 🔇 Suppress duplicate skill load announcements — if same skill already announced, don't repeat.
+- 📦 Compact mode: when ≥2 skills loaded, use: `[⚡PikaKit] L2/@{skill-a} + @{skill-b} → RUNNING`
+
+---
+
+## ⛔ HARD GATE — Mandatory Pre-Code Check
+
+> 🔴 **THIS IS NOT OPTIONAL.** Every code response MUST pass this gate.
+> Agent that skips this gate = **PROTOCOL VIOLATION** = user loses trust.
+
+### Before writing ANY code, agent MUST produce this header:
+
+```text
+[⚡PikaKit] L{0-3}/@{skill-name} → INIT → RUNNING
+```
+
+### Gate Rules:
+
+| Level | Required Actions | Failure = |
+|-------|-----------------|-----------|
+| **L0** (Question) | Answer directly, no gate needed | — |
+| **L1** (Quick fix, <10 lines) | Classify + `[⚡PikaKit] L1/@{skill} → INIT` | **INVALID response** |
+| **L2** (Multi-file) | Classify + `view_file SKILL.md` + `[⚡PikaKit] L2/@{skill}` | **INVALID response** |
+| **L3** (Architecture) | Classify + `view_file SKILL.md + AGENTS.md` + Plan | **INVALID response** |
+
+### Self-Check Trigger:
+
+```
+EVERY TIME you are about to write code or create files:
+  → "Did I output the Task Level + Skill header?"
+  → If NO → STOP. Output header FIRST. Then continue.
+  → If YES → Proceed with code.
+```
+
+### No Exceptions:
+
+- ❌ "I'll do it next time" → NO. Do it NOW.
+- ❌ "The task is a continuation" → Still requires header.
+- ❌ "Context was truncated/resumed" → Still requires header.
+- ❌ "It's obvious which skill" → Still requires `view_file` for L2+.
+
+> **Why this exists:** Without enforcement, agents skip skill routing 100% of the time.
+> This gate converts a "suggestion" into a **hard requirement**.
+
+---