Phase 0 Task Prompts — Index
This directory contains agent-ready prompts for every Phase 0 task. Each prompt is self-contained — give it to an LLM agent (Claude, GPT, etc.) cold and the agent has everything it needs to ship that task.
How to Use This Directory
- Start here: Read agent-bootstrap.md — the master bootstrap prompt for all agents.
- Understand the sequence: Read first-7-prs.md — the critical-path ordering (7 PRs to working server).
- Pick a task: Look at the tables below. Prefer unblocked tasks at the top of the dependency graph.
- Copy the prompt: Open the relevant file (e.g.,
p0.1-infrastructure.mdfor P0.1 tasks). - Find “Ready-to-paste agent prompt”: Copy the full prompt block and give it to your agent.
- Verify before merge: Use the per-task verification checklist before you land the PR.
Phase 0 Task Index
Group Summary
| Group | File | Sub-tasks | Total Effort | Unblocks |
|---|---|---|---|---|
| P0.1 — Infrastructure | p0.1-infrastructure.md | P0.1.1, P0.1.2, P0.1.3, P0.1.4 | 2–3d | P0.2.x, P0.3.x, P0.4.x |
| P0.2 — Core System | p0.2-alpha-system-core.md | P0.2.1, P0.2.2, P0.2.3, P0.2.4 | 3–4d | P0.3.x, P0.5.x, P0.7.x, P0.8.x |
| P0.3 — Task Pipeline | p0.3-beta-task-pipeline.md | P0.3.1, P0.3.2, P0.3.3, P0.3.4 | 2–3d | P0.4.x, P0.6.x, P0.9.x |
| P0.4 — Server Lifecycle | p0.4-gamma-lifecycle.md | P0.4.1, P0.4.2 | 1–2d | P0.5.x, P0.6.x |
| P0.5 — Model Router | p0.5-delta-router.md | P0.5.1, P0.5.2 | 2–3d | P0.6.x, P0.7.x |
| P0.6 — Skill Registry | p0.6-epsilon-skills.md | P0.6.1, P0.6.2, P0.6.3 | 2–3d | P0.7.x, P0.9.x |
| P0.7 — Decision Trail | p0.7-zeta-decision-trail.md | P0.7.1, P0.7.2, P0.7.3 | 2–3d | P0.8.x, P0.9.x |
| P0.8 — Proof Store | p0.8-eta-proof-store.md | P0.8.1, P0.8.2, P0.8.3 | 2–3d | P0.9.x |
| P0.9 — Integrations | p0.9-nu-integrations.md | P0.9.1, P0.9.2, P0.9.3 | 1–2d | (Phase 1+) |
Per-Task Index (All 28 Tasks)
| Task ID | Title | Effort | Depends on | Status | File |
|---|---|---|---|---|---|
| P0.1.1 | Package Setup | S | — | spec-ready | p0.1-infrastructure.md |
| P0.1.2 | Test Runner + Linter | S | P0.1.1 | spec-ready | p0.1-infrastructure.md |
| P0.1.3 | CI Pipeline | S | P0.1.2 | spec-ready | p0.1-infrastructure.md |
| P0.1.4 | Environment Validation | S | P0.1.1 | spec-ready | p0.1-infrastructure.md |
| P0.2.1 | MCP Server Bootstrap | M | P0.1.2 | spec-ready | p0.2-alpha-system-core.md |
| P0.2.2 | SQLite Initialization | L | P0.1.4 | spec-ready | p0.2-alpha-system-core.md |
| P0.2.3 | Two-Phase Startup | M | P0.2.1, P0.2.2 | spec-ready | p0.2-alpha-system-core.md |
| P0.2.4 | Middleware Stack | M | P0.2.3 | spec-ready | p0.2-alpha-system-core.md |
| P0.3.1 | GSD State Machine | S | P0.2.2 | spec-ready | p0.3-beta-task-pipeline.md |
| P0.3.2 | Task CRUD Routes | S | P0.3.1 | spec-ready | p0.3-beta-task-pipeline.md |
| P0.3.3 | Validation + Errors | S | P0.3.2 | spec-ready | p0.3-beta-task-pipeline.md |
| P0.3.4 | Tool Binding | S | P0.3.3 | spec-ready | p0.3-beta-task-pipeline.md |
| P0.4.1 | Boot Watcher | S | P0.4 | spec-ready | p0.4-gamma-lifecycle.md |
| P0.4.2 | Shutdown Handler | S | P0.4.1 | spec-ready | p0.4-gamma-lifecycle.md |
| P0.5.1 | Intent Scoring | M | P0.2.3 | spec-ready | p0.5-delta-router.md |
| P0.5.2 | Fallback Chain | M | P0.5.1 | spec-ready | p0.5-delta-router.md |
| P0.6.1 | Skill Loader | M | P0.3.2 | spec-ready | p0.6-epsilon-skills.md |
| P0.6.2 | Dispatcher | M | P0.6.1 | spec-ready | p0.6-epsilon-skills.md |
| P0.6.3 | Lifecycle Hooks | M | P0.6.2 | spec-ready | p0.6-epsilon-skills.md |
| P0.7.1 | Thought Record Creation | M | P0.2.3 | spec-ready | p0.7-zeta-decision-trail.md |
| P0.7.2 | Hash Chaining | M | P0.7.1 | spec-ready | p0.7-zeta-decision-trail.md |
| P0.7.3 | Audit Logging | M | P0.7.2 | spec-ready | p0.7-zeta-decision-trail.md |
| P0.8.1 | Merkle Tree Builder | L | P0.2.2 | spec-ready | p0.8-eta-proof-store.md |
| P0.8.2 | Inclusion Proofs | L | P0.8.1 | spec-ready | p0.8-eta-proof-store.md |
| P0.8.3 | Retention Zones | M | P0.8.2 | spec-ready | p0.8-eta-proof-store.md |
| P0.9.1 | GitHub Integration | S | P0.3.4 | spec-ready | p0.9-nu-integrations.md |
| P0.9.2 | Obsidian Bridge | S | P0.3.4 | spec-ready | p0.9-nu-integrations.md |
| P0.9.3 | Claude API Router | M | P0.5.2 | spec-ready | p0.9-nu-integrations.md |
Critical-Path Summary
The 7 PRs that transform Colibri from empty repo to working MCP server:
-
P0.1.1 — Package Setup
→ Unblocks: P0.1.2, P0.1.4 -
P0.1.2 — Test Runner + Linter
→ Unblocks: P0.1.3, P0.2.1 -
P0.1.4 — Environment Validation
→ Unblocks: P0.2.2 -
P0.2.1 — MCP Server Bootstrap
→ Unblocks: P0.2.3, P0.3.4, P0.4.4 -
P0.2.2 — SQLite Initialization
→ Unblocks: P0.2.3, P0.3.1 -
P0.2.3 — Two-Phase Startup
→ Unblocks: P0.2.4, P0.3.4 -
P0.3.1 — GSD State Machine
→ Unblocks: P0.3.2, P0.3.4
After PR-7 merges: Colibri is bootstrapped. Full dogfooding mode enabled.
Parallel-Track Summary
After each critical-path unblock, these tasks can run in parallel:
-
After PR-1 (P0.1.1 merges):
P0.1.2 (linter setup) can start immediately -
After PR-2 (P0.1.2 merges):
P0.1.3 (CI) + P0.2.1 (server bootstrap) can start in parallel -
After PR-3 (P0.1.4 merges):
P0.2.2 (SQLite) can start -
After PR-4 (P0.2.1 merges):
P0.2.3 (two-phase startup) + P0.3.4 (tool binding) + P0.4.1 (boot watcher) can start in parallel -
After PR-5 (P0.2.2 merges):
P0.2.3 (two-phase startup) + P0.3.1 (GSD) can start in parallel -
After PR-6 (P0.2.3 merges):
P0.2.4 (middleware) + P0.3.4 (tool binding) can start in parallel -
After PR-7 (P0.3.1 merges):
All remaining P0 tasks can start in parallel:
P0.3.2, P0.3.3, P0.3.4 (task pipeline)
P0.4.2 (shutdown handler)
P0.5.1, P0.5.2 (model router)
P0.6.1, P0.6.2, P0.6.3 (skill registry)
P0.7.1, P0.7.2, P0.7.3 (decision trail)
P0.8.1, P0.8.2, P0.8.3 (proof store)
P0.9.1, P0.9.2, P0.9.3 (integrations)
Phase 1 Task Prompts
| Group | File | Sub-tasks | Total Effort | Status |
|---|---|---|---|---|
| P1.1–P1.5 — κ Rule Engine | p1.1-kappa-rule-engine.md | P1.1.1–P1.5.5 (20 tasks) | ~4–6w | planning (R76) |
20-sub-task breakdown: P1.1 Integer Math (3) · P1.2 DSL Parser (4) · P1.3 Deterministic Interpreter (4) · P1.4 Admission Layer (4) · P1.5 Governance / Rule Versioning (5). Full canonical spec at task-breakdown.md §Phase 1. Phase 1 implementation officially starts at R81 per roadmap.md.
After Phase 0
Once all 28 Phase 0 tasks are done:
- Phase 1: p1.1-kappa-rule-engine.md — κ Rule Engine (integer math, DSL parser, interpreter, admission layer, governance / rule versioning). 20 sub-tasks. Ready-to-paste prompts shipped R76.P1 (2026-04-18).
- Phase 2+: See task-breakdown.md for the full roadmap (9 total phases)
Phase 1.5 — δ Multi-Model Router (R91+, planned)
Once Phase 1 κ Rule Engine ships and the four
ADR-005 §Implementation
trigger conditions hold, the prompts below graduate δ from Phase 0 library
stubs to multi-model routing. Scope covers real 7-dimension scoring,
3 non-Claude adapters, N-member fallback with circuit breaker, cost
accounting, 4 router_* MCP tools, cross-model parity tests, candidate
table population, and ζ decision-trail integration.
| Group | File | Sub-tasks |
|---|---|---|
| P1.5 — δ Router Graduation | p1.5-delta-router-graduation.md | P1.5.1 – P1.5.10 |
See Also
- agent-bootstrap.md — Master bootstrap prompt (read FIRST)
- first-7-prs.md — Critical-path PR sequence with full context
- agent-handoff-protocol.md — Multi-agent handoff spec
- task-breakdown.md — Canonical 63-task breakdown (all phases)
- PHASE-0-EXECUTION-GUIDE.md — Phase 0 entry point