antigravity-ai-kit 2.1.0 → 3.0.1

package/.agent/workflows/orchestrate.md

@@ -1,114 +1,171 @@
----
-description: Multi-agent orchestration for complex tasks requiring multiple specialists.
----
-
-# /orchestrate - Multi-Agent Coordination
-
-$ARGUMENTS
-
----
-
-## 🔴 CRITICAL: 2-Phase Orchestration
-
-### PHASE 1: PLANNING (Sequential)
-
-| Step | Agent            | Action                         |
-| ---- | ---------------- | ------------------------------ |
-| 1    | `planner`        | Create implementation plan     |
-| 2    | `explorer-agent` | Codebase discovery (if needed) |
-
-> 🔴 **NO OTHER AGENTS during planning!**
-
-### ⏸️ CHECKPOINT: User Approval
-
-After plan is complete, ASK:
-
-```
-✅ Plan created: docs/PLAN.md
-
-Do you approve? (Y/N)
-- Y: Start implementation
-- N: I'll revise the plan
-```
-
-> 🔴 **DO NOT proceed to Phase 2 without explicit user approval!**
-
-### PHASE 2: IMPLEMENTATION (Parallel after approval)
-
-| Group      | Agents                                    |
-| ---------- | ----------------------------------------- |
-| Foundation | `database-architect`, `security-reviewer` |
-| Core       | `architect`, `mobile-developer`           |
-| Polish     | `tdd-guide`, `devops-engineer`            |
-
----
-
-## Agent Selection Matrix
-
-| Domain      | Keywords                              | Agent(s)                  |
-| ----------- | ------------------------------------- | ------------------------- |
-| Security    | "security", "auth", "vulnerabilities" | `security-reviewer`       |
-| Backend     | "API", "database", "server"           | `database-architect`      |
-| Frontend    | "UI", "component", "page"             | `architect`               |
-| Mobile      | "mobile", "expo", "react native"      | `mobile-developer`        |
-| Testing     | "test", "coverage", "e2e"             | `tdd-guide`, `e2e-runner` |
-| DevOps      | "deploy", "CI/CD", "production"       | `devops-engineer`         |
-| Performance | "slow", "optimize", "speed"           | `performance-optimizer`   |
-
----
-
-## Orchestration Protocol
-
-1. **Analyze Task Domains** — Identify all domains involved
-2. **Phase Detection** — Plan exists? User approved?
-3. **Execute** — Planning or implementation based on phase
-4. **Context Passing** — Pass full context to subagents
-5. **Verification** — Run quality gates
-6. **Synthesize** — Combine outputs into report
-
----
-
-## Context Passing (MANDATORY)
-
-When invoking ANY subagent, include:
-
-1. **Original User Request:** Full text
-2. **Decisions Made:** All answers to Socratic questions
-3. **Previous Agent Work:** Summary of completed work
-4. **Current Plan State:** If plan exists
-
----
-
-## Output Format
-
-```markdown
-## 🎭 Orchestration Complete
-
-### Agents Invoked
-
-- `architect` → [summary]
-- `database-architect` → [summary]
-- `tdd-guide` → [summary]
-
-### Deliverables
-
-- [list of created/modified files]
-
-### Verification
-
-- [test results, lint results]
-
-### Next Steps
-
-- [suggestions for follow-up]
-```
-
----
-
-## Examples
-
-```
-/orchestrate build e-commerce platform
-/orchestrate add user authentication
-/orchestrate refactor database layer
-```
+---
+description: Multi-agent orchestration for complex tasks requiring multiple specialists.
+version: 2.1.0
+sdlc-phase: reactive
+agents: [planner, explorer-agent]
+skills: [parallel-agents, intelligent-routing]
+commit-types: [feat, refactor, fix]
+---
+
+# /orchestrate — Multi-Agent Coordination
+
+> **Trigger**: `/orchestrate [task description]`
+> **Lifecycle**: Reactive — complex multi-domain tasks at any SDLC phase
+
+> [!CAUTION]
+> This workflow coordinates multiple agents on the same codebase. Ensure clear domain boundaries to avoid conflicting changes. Phase 2 (implementation) requires explicit user approval.
+
+> [!TIP]
+> This workflow leverages the **parallel-agents** and **intelligent-routing** skills. Read `.agent/skills/parallel-agents/SKILL.md` and `.agent/skills/intelligent-routing/SKILL.md` for extended guidance.
+
+---
+
+## Critical Rules
+
+1. **2-Phase protocol** — always plan before implementing; never skip Phase 1
+2. **User approval gate** — Phase 2 cannot start without explicit user approval
+3. **Context passing mandatory** — every subagent must receive full context (original request, decisions made, previous work)
+4. **No agent conflicts** — agents must work on separate files or clearly delineated domains
+5. **Verification required** — run quality gates after all agents complete
+
+---
+
+## Steps
+
+### PHASE 1: Planning (Sequential)
+
+// turbo
+1. **Analyze Task Domains**
+   - Identify all domains involved (backend, frontend, database, security, etc.)
+   - Map which agents are needed
+
+// turbo
+2. **Create Orchestration Plan**
+   - Use `planner` agent for structured task breakdown
+   - Use `explorer-agent` for codebase discovery if needed
+   - Define execution order and dependencies
+
+> 🔴 **NO other agents during Phase 1!**
+
+3. **Checkpoint: User Approval**
+
+   ```
+   ✅ Orchestration plan created.
+
+   Agents needed: [list]
+   Estimated scope: [file count]
+
+   Approve to start implementation? (Y/N)
+   ```
+
+> 🔴 **DO NOT proceed to Phase 2 without explicit user approval!**
+
+### PHASE 2: Implementation (After Approval)
+
+4. **Execute in Groups**
+
+   | Group | Agents | Domain |
+   | :--------- | :------------------------------- | :------------------------- |
+   | Foundation | `database-architect`, `security-reviewer` | Data layer, security |
+   | Core | `architect`, `backend-specialist`, `frontend-specialist`, `mobile-developer` | Application logic |
+   | Quality | `tdd-guide`, `e2e-runner` | Test coverage |
+   | Operations | `devops-engineer`, `reliability-engineer` | Infrastructure, reliability |
+
+5. **Context Passing** (for every subagent invocation)
+   - Original user request (full text)
+   - All decisions from Socratic questions
+   - Summary of previous agent work
+   - Current plan state
+
+// turbo
+6. **Verification**
+   - Run full test suite
+   - Run lint and type-check
+   - Verify build succeeds
+   - Synthesize results into report
+
+---
+
+## Agent Selection Matrix
+
+| Domain | Keywords | Agent(s) |
+| :----------- | :------------------------------------ | :------------------------------------------------------ |
+| Architecture | "design", "structure", "pattern" | `architect`, `planner` |
+| Backend | "API", "database", "server" | `backend-specialist`, `database-architect` |
+| Frontend | "UI", "component", "page" | `frontend-specialist` |
+| Mobile | "mobile", "expo", "react native" | `mobile-developer` |
+| Security | "security", "auth", "vulnerabilities" | `security-reviewer` |
+| Testing | "test", "coverage", "e2e" | `tdd-guide`, `e2e-runner` |
+| DevOps | "deploy", "CI/CD", "production" | `devops-engineer` |
+| Performance | "slow", "optimize", "speed" | `performance-optimizer` |
+| Reliability | "uptime", "monitoring", "resilience" | `reliability-engineer` |
+| Code Quality | "refactor", "clean", "lint" | `refactor-cleaner`, `code-reviewer` |
+
+---
+
+## Output Template
+
+```markdown
+## 🎭 Orchestration Complete
+
+### Agents Invoked
+
+| Agent | Domain | Summary |
+| :---- | :----- | :------ |
+| `[agent]` | [domain] | [what was done] |
+
+### Deliverables
+
+| Action | File | Agent |
+| :----- | :--- | :---- |
+| Created | `path/to/file` | [agent] |
+| Modified | `path/to/file` | [agent] |
+
+### Verification
+
+- ✅ Tests: [N] passing
+- ✅ Build: successful
+- ✅ Lint: clean
+
+### Next Steps
+
+- [suggestions for follow-up workflows]
+```
+
+---
+
+## Governance
+
+**PROHIBITED:**
+- Skipping Phase 1 (planning) and jumping to implementation
+- Starting Phase 2 without explicit user approval
+- Invoking agents without passing full context
+- Allowing agents to work on overlapping files without coordination
+- Skipping verification after agent work completes
+- Skipping failed steps · proceeding without resolution
+
+**REQUIRED:**
+- 2-Phase protocol (plan → approve → implement)
+- Full context passing to every subagent
+- Domain boundary enforcement
+- Quality verification after all agents complete
+
+---
+
+## Completion Criteria
+
+- [ ] Task domains are analyzed and agents are selected
+- [ ] Phase 1 plan is created and presented
+- [ ] User has explicitly approved Phase 2
+- [ ] All agents have completed their work
+- [ ] Context was passed to every subagent invocation
+- [ ] Verification passes (tests, build, lint)
+- [ ] Final synthesized report is delivered
+
+---
+
+## Related Resources
+
+- **Cross-cutting**: Can be invoked at any SDLC phase for complex tasks
+- **Skills**: `.agent/skills/parallel-agents/SKILL.md` · `.agent/skills/intelligent-routing/SKILL.md`
+- **Agents**: See `.agent/agents/` for full agent catalog

package/.agent/workflows/plan.md

@@ -1,72 +1,147 @@
----
-description: Create implementation plan. Invokes planner agent for structured task breakdown.
----
-
-# /plan - Implementation Planning
-
-$ARGUMENTS
-
----
-
-## 🔴 CRITICAL RULES
-
-1. **NO CODE WRITING** — This creates plan only
-2. **Use planner agent** — Invoke for structured breakdown
-3. **Socratic Gate** — Ask clarifying questions first
-4. **Dynamic Naming** — Name based on task
-
----
-
-## Task
-
-Create implementation plan using the `planner` agent.
-
-### Steps
-
-1. **Clarify Requirements**
-   - Ask 3 questions minimum (Socratic Gate)
-   - Purpose, scope, constraints
-
-2. **Create Plan**
-   - Task breakdown with verification criteria
-   - Agent assignments if multi-domain
-   - Clear done criteria
-
-3. **Output Location**
-   - Plan in project root or `docs/` folder
-   - Named dynamically: `PLAN-{task-slug}.md`
-
----
-
-## Naming Examples
-
-| Request                 | Plan File                |
-| ----------------------- | ------------------------ |
-| `/plan e-commerce cart` | `PLAN-ecommerce-cart.md` |
-| `/plan mobile app`      | `PLAN-mobile-app.md`     |
-| `/plan dark mode`       | `PLAN-dark-mode.md`      |
-| `/plan auth fix`        | `PLAN-auth-fix.md`       |
-
----
-
-## After Planning
-
-```
-✅ Plan created: docs/PLAN-{slug}.md
-
-Next steps:
-- Review the plan
-- Approve to start implementation
-- Or modify plan manually
-```
-
----
-
-## Examples
-
-```
-/plan e-commerce site with cart
-/plan mobile app for fitness tracking
-/plan SaaS dashboard with analytics
-/plan add authentication
-```
+---
+description: Create implementation plan. Invokes planner agent for structured task breakdown.
+version: 2.1.0
+sdlc-phase: plan
+agents: [planner]
+skills: [plan-writing, brainstorming]
+commit-types: [docs]
+---
+
+# /plan — Implementation Planning
+
+> **Trigger**: `/plan [task description]`
+> **Lifecycle**: Plan — first step of SDLC after discovery
+
+> [!IMPORTANT]
+> This workflow creates plans, NOT code. No implementation during planning. All plans require user approval before execution begins.
+
+> [!TIP]
+> This workflow leverages the **plan-writing** skill. Read `.agent/skills/plan-writing/SKILL.md` for extended guidance.
+
+---
+
+## Critical Rules
+
+1. **No code writing** — this workflow produces plans only
+2. **Socratic gate** — ask at least 3 clarifying questions before creating a plan
+3. **Dynamic naming** — name plan files based on the task (e.g., `PLAN-auth-fix.md`)
+4. **Verification criteria** — every task in the plan must have clear "done" criteria
+5. **User approval required** — never proceed to implementation without explicit approval
+6. **Small, focused tasks** — break down work into atomic, verifiable steps
+
+---
+
+## Argument Parsing
+
+| Command | Action |
+| :----------------------- | :---------------------------------------------- |
+| `/plan` | Prompt for task description |
+| `/plan [description]` | Create implementation plan for the described task |
+
+---
+
+## Steps
+
+// turbo
+1. **Clarify Requirements** (Socratic Gate)
+   - Ask at least 3 clarifying questions about purpose, scope, and constraints
+   - Confirm acceptance criteria and edge cases
+   - Identify relevant existing code and patterns
+
+// turbo
+2. **Explore Codebase**
+   - Scan project structure and architecture
+   - Identify files, modules, and patterns relevant to the task
+   - Note dependencies and integration points
+
+3. **Create Plan**
+   - Break down the task into small, focused steps
+   - Assign verification criteria to each step
+   - Order tasks logically (dependencies first)
+   - Identify which agents are needed for multi-domain tasks
+   - Save plan to `docs/PLAN-{task-slug}.md`
+
+4. **Present for Approval**
+   - Show the plan summary to the user
+   - Wait for explicit approval before any implementation
+
+---
+
+## Naming Convention
+
+| Request | Plan File |
+| :----------------------- | :--------------------------- |
+| `/plan e-commerce cart` | `PLAN-ecommerce-cart.md` |
+| `/plan mobile app` | `PLAN-mobile-app.md` |
+| `/plan auth fix` | `PLAN-auth-fix.md` |
+
+---
+
+## Output Template
+
+```markdown
+## 📋 Plan: [Task Name]
+
+### Scope
+
+[What this plan covers and what it doesn't]
+
+### Tasks
+
+1. [ ] [Task description] — **Verify**: [done criteria]
+2. [ ] [Task description] — **Verify**: [done criteria]
+3. [ ] [Task description] — **Verify**: [done criteria]
+
+### Agent Assignments (if multi-domain)
+
+| Task | Agent | Domain |
+| :--- | :---- | :----- |
+| [task] | [agent] | [domain] |
+
+### Risks & Considerations
+
+- [risk or constraint]
+
+---
+
+✅ Plan saved: `docs/PLAN-{slug}.md`
+
+Approve to start implementation with `/create` or `/enhance`.
+```
+
+---
+
+## Governance
+
+**PROHIBITED:**
+- Writing implementation code during planning
+- Proceeding to implementation without user approval
+- Creating vague, unverifiable tasks
+- Skipping the Socratic gate
+- Skipping failed steps · proceeding without resolution
+
+**REQUIRED:**
+- At least 3 clarifying questions before planning
+- Verification criteria for every task
+- User approval before implementation begins
+- Plan file saved in `docs/` with dynamic name
+
+---
+
+## Completion Criteria
+
+- [ ] Clarifying questions asked and answered
+- [ ] Codebase explored for relevant context
+- [ ] Plan created with verifiable tasks
+- [ ] Plan saved to `docs/PLAN-{slug}.md`
+- [ ] User has reviewed and approved the plan
+- [ ] After approval: proceed to `/create` or `/enhance` for implementation
+
+---
+
+## Related Resources
+
+- **Previous**: `/brainstorm` (explore options) · `/quality-gate` (validate approach)
+- **Next**: `/create` (scaffold new features) · `/enhance` (iterative development)
+- **Skill**: `.agent/skills/plan-writing/SKILL.md`
+- **Agent**: `planner` agent (see `.agent/agents/planner.md`)