Prompt Pipeline

Pipeline Flow:
  refs/ ──> [001] ──> specs/ ──> [002] ──> plans/ ──> [003] ──> src/ + tests/
                                   ^                     |
                                   |                     |
                                   +── impl/ <───────────+
                                   (bidirectional flow)

# 001: Generate Specs from Reference Materials

## Runtime Inputs
- Framework: {FRAMEWORK}
- Build command: {BUILD_COMMAND}
- Test command: {TEST_COMMAND}

## Context
Read all files in `context/refs/`. These are the source of truth.

## Task
Decompose the reference materials into domain-specific specifications:
1. Create `context/kits/cavekit-overview.md` as the index file
2. Create one `context/kits/spec-{domain}.md` per domain
3. Each spec must include: Scope, Requirements with Acceptance Criteria, Dependencies, Out of Scope, Cross-References

## Exit Criteria
- [ ] All domains from reference materials have corresponding spec files
- [ ] Every requirement has at least one testable acceptance criterion
- [ ] cavekit-overview.md indexes all spec files with one-line summaries
- [ ] Cross-references link related specs

## Completion Signal
<all-tasks-complete>

Pipeline Flow:
  old-code ──> [001] ──> reference/ ──> [002] ──> specs/ ──> [003] ──> validated specs
                                                                            |
       +────────────────────────────────────────────────────────────────────+
       |
       v
  specs/ ──> [004] ──> plans/ ──> [005] ──> src/ + tests/ ──> [006] ──> updated specs
                                                                            |
       +────────────────────────────────────────────────────────────────────+
       |
       v
  (loop back to 002 for refinement)

Prompt 002 (Plans):
  READS:  context/kits/     (what to build)
  READS:  context/impl/      (what has been built, what failed)
  WRITES: context/plans/     (how to build it)

Prompt 003 (Implement):
  READS:  context/plans/     (how to build it)
  READS:  context/kits/     (acceptance criteria)
  WRITES: src/, tests/       (the code)
  WRITES: context/impl/      (progress tracking)
  WRITES: context/plans/     (updates to plans based on implementation reality)

## Runtime Inputs
- Framework: {FRAMEWORK}           # e.g., "React + Vite", "Tauri + Svelte"
- Build command: {BUILD_COMMAND}   # e.g., "npm run build", "cargo build"
- Test command: {TEST_COMMAND}     # e.g., "npm test", "pytest"
- Lint command: {LINT_COMMAND}     # e.g., "npm run lint", "cargo clippy"
- Source dir: {SRC_DIR}            # e.g., "src/", "lib/"
- Test dir: {TEST_DIR}            # e.g., "tests/", "__tests__/"

Agent Team Structure:
  Lead (delegate mode -- never writes code directly)
  +-- Teammate A: domain-auth
  |   Owns: src/auth/*, context/impl/impl-auth.md
  |   Dispatch: Agent tool
  +-- Teammate B: domain-data
  |   Owns: src/data/*, context/impl/impl-data.md
  |   Dispatch: Agent tool
  +-- Teammate C: domain-ui
      Owns: src/ui/*, context/impl/impl-ui.md
      Dispatch: Agent tool

Execution Timeline:
  Batch 1: [Teammate A] [Teammate B] [Teammate C]
           ─────────────────────────────────────────> complete, shutdown
  Batch 2: [Teammate D] [Teammate E] [Teammate F]
           ─────────────────────────────────────────> complete, shutdown

## File Ownership
| File/Pattern | Owner |
|-------------|-------|
| `src/auth/**` | domain-auth |
| `src/data/**` | domain-data |
| `src/ui/**` | domain-ui |
| `src/shared/types.ts` | domain-data |
| `context/impl/impl-auth.md` | domain-auth |

## Exit Criteria
- [ ] All T- tasks are DONE or documented as BLOCKED
- [ ] {BUILD_COMMAND} passes with zero errors
- [ ] {TEST_COMMAND} passes with zero failures
- [ ] All modified source files have corresponding test coverage
- [ ] context/impl/ updated with current status

## Completion Signal
When ALL exit criteria are met, output exactly:
<all-tasks-complete>

This signal is used by the iteration loop to detect when to stop.

## Spawn Template for Teammate

You are implementing {DOMAIN} for the {PROJECT_NAME} project.

### Your Role
- You own: {FILE_PATTERNS}
- Dispatched via the Agent tool
- Your impl tracking: context/impl/impl-{DOMAIN}.md

### Context to Read First
1. context/kits/spec-{DOMAIN}.md (WHAT to build)
2. context/plans/plan-{DOMAIN}.md (HOW to build it)
3. context/impl/impl-{DOMAIN}.md (what has been done)
4. git log --oneline -20 (recent history)

### Task
{TASK_DESCRIPTION}

### Exit Criteria
- [ ] {CRITERIA}

### Halting Conditions
- Do NOT push to remote unless explicitly asked
- Do NOT modify files outside your ownership
- If blocked for more than 20 minutes, document the blocker and stop

## Halting Conditions
- Do NOT push to remote unless explicitly asked
- Do NOT modify files outside your file ownership table
- Do NOT delete test files or skip failing tests
- If a task takes more than 20 minutes, document findings and move on
- If you encounter a circular dependency, document it and stop
- Commit frequently to preserve progress

## When to Use Sub-Agents
- Reading large documentation files
- Running and parsing test output
- Generating boilerplate code
- Researching framework APIs
- Performing file-by-file migrations

## Sub-Agent Rules
- Max 3 concurrent sub-agents per teammate
- Each sub-agent gets a focused, self-contained task
- Sub-agent results are summarized back to the teammate
- Sub-agents do NOT inherit the teammate's conversation history