planforge-step1-preflight-check

name	planforge-step1-preflight-check
description	Pipeline Step 1 — Pre-flight checks before plan hardening. Verifies git state, roadmap link, plan file, guardrails, and domain-specific instruction files.
metadata	{"author":"plan-forge","source":".github/prompts/step1-preflight-check.prompt.md"}

---

description: "Pipeline Step 1 — Pre-flight checks before plan hardening. Verifies git state, roadmap link, plan file, guardrails, and domain-specific instruction files."

Step 1: Pre-flight Check

Pipeline: Step 1 of 5 (Session 1 — Plan Hardening)
When: Before hardening any *-PLAN.md
Model suggestion: Any model / Copilot Auto (10% token savings) — checklist verification works well on all models
Next Step: step2-harden-plan.prompt.md

Replace <YOUR-PLAN> with your plan filename (without path or .md extension).

---

Act as a PRE-FLIGHT CHECK AGENT for plan hardening.

Run these checks and report results. If any check fails, report the failure and do not proceed to Step 2.

1. GIT STATE — Run git pull origin main and git status. Report: clean / dirty (list uncommitted files if dirty).

2. ROADMAP LINK — Read docs/plans/DEPLOYMENT-ROADMAP.md. Confirm the phase for exists with a one-line goal. Report: ✅ found (quote the goal) / ❌ missing.

3. PLAN FILE — Confirm docs/plans/.md exists and is non-empty. Report: ✅ exists (N lines) / ❌ not found.

4. CORE GUARDRAILS — Confirm these files exist and are non-empty:

   • .github/copilot-instructions.md
   • .github/instructions/architecture-principles.instructions.md
   • AGENTS.md Report: ✅ all present / ❌ missing (list which).

4b. AGENTIC FILES — Check if prompt templates, agent definitions, and skills exist:

• .github/prompts/ — list *.prompt.md files found (0 is OK for non-preset repos)
• .github/agents/ — list *.agent.md files found
• .github/skills/ — list */SKILL.md files found Report: ✅ N prompts, N agents, N skills found / ⚠️ none found (optional — won't block)

5. DOMAIN GUARDRAILS — Scan .md for keywords to identify relevant domains. For each domain detected, confirm the matching guardrail file exists:
   • UI/Component/Frontend/Razor/React/Vue → .github/instructions/frontend.instructions.md (or blazor/react specific)
   • Database/SQL/Repository/ORM/migration → .github/instructions/database.instructions.md
   • API/Route/Controller/REST → .github/instructions/api-patterns.instructions.md
   • Auth/OAuth/JWT/OIDC/session → .github/instructions/auth.instructions.md
   • GraphQL/Schema/Resolver → .github/instructions/graphql.instructions.md
   • Security/CORS/Secrets/Validation → .github/instructions/security.instructions.md
   • Docker/K8s/deploy/CI → .github/instructions/deploy.instructions.md
   • Test/spec/coverage → .github/instructions/testing.instructions.md Report: domains detected + guardrail status for each.

Output a summary table:

Check	Result	Details
Git state	✅/❌	...
Roadmap link	✅/❌	...
Plan file	✅/❌	...
Core guardrails	✅/❌	...
Agentic files	✅/⚠️	...
Domain guardrails	✅/❌	...

If ALL pass: "Pre-flight complete ✅ — proceed to Step 2 (Harden the Plan)" If ANY fail: "Pre-flight FAILED ❌" + list exactly what to fix.

Tip: For deeper diagnostics (environment tools, VS Code config, version currency, common problems), suggest running pforge smith.

---

Persistent Memory (if OpenBrain is configured)

• Before checking: search_thoughts("preflight blockers", project: "TimeTracker", created_by: "copilot-vscode", type: "bug") — check if prior preflights failed for known reasons
• If preflight fails: capture_thought("Preflight blocker: <what failed and why>", project: "TimeTracker", created_by: "copilot-vscode", source: "plan-forge-step-1", type: "bug") — persist the blocker so it's caught earlier next time
• If preflight passes with notable decisions: capture_thought("Preflight passed: <key confirmations — branch strategy, resolved ambiguities>", project: "TimeTracker", created_by: "copilot-vscode", source: "plan-forge-step-1", type: "decision") — persist confirmations so next phase doesn't re-ask