/aif-status — Project Setup Status & Next Steps

You are checking the health of an aiframework-configured project and guiding the user on what to do next.

Step 0: Telemetry (silent, non-blocking)

aiframework-telemetry skill_invoked skill=aif-status 2>/dev/null || true

Steps

Step 1: Check Generated Files

Read and verify these files exist. For each, note if it's present, its line count, and whether it looks correct:

CLAUDE.md                           — Project brain (should be 80-150 lines)
AGENTS.md                           — Universal agent config (should be 30-80 lines)
.cursorrules                        — Cursor IDE config
.claude/rules/workflow.md           — Workflow rules
.claude/rules/pipeline.md           — 12-stage pipeline (moderate+ only)
.claude/rules/session-protocol.md   — Session checklists (moderate+ only)
.claude/rules/invariants.md         — Domain invariants (moderate+ only)
.claude/settings.json               — Claude Code permissions
.claude/skills/                     — Custom skills
.githooks/pre-push                  — Quality gate hook
.github/workflows/ci.yml            — CI pipeline
docs/reference/architecture.md      — Architecture reference (moderate+ only)
vault/memory/status.md              — Session memory
tools/learnings/                    — Learning capture
CHANGELOG.md                        — Release history
VERSION                             — Version tracking

Use Glob to check: **/{CLAUDE.md,AGENTS.md,.cursorrules} and .claude/rules/*.md and .claude/skills/*/SKILL.md

Step 2: Check Manifest Quality

Read .aiframework/manifest.json and evaluate:

1. Language: Is it correct? Does it match the actual primary language?
2. Framework: Detected or "none"?
3. Archetype: Does it make sense for this project?
4. Commands: Are build/test/lint configured or NOT_CONFIGURED?
5. Domains: Any detected? Are they accurate (not false positives)?
6. Description: Is it a real description or HTML/empty?
7. Complexity: simple/moderate/complex/enterprise — affects tier

Step 3: Check Code Index

Read .aiframework/code-index.json metadata:

jq '._meta' .aiframework/code-index.json

Report: files indexed, symbols extracted, modules found. If 0 files: code index wasn't run (use aiframework index).

Step 4: Check Vault Health

vault/.vault/scripts/vault-tools.sh doctor 2>/dev/null || echo "Vault not initialized"

Step 5: Check for Enhancement Opportunities

Identify gaps that /aif-enhance can fill:

• Framework detected but no framework-specific invariants? → /aif-enhance
• Commands NOT_CONFIGURED for build/test/lint? → User needs to add to package.json/Makefile
• No domains detected? → Project may need /aif-enhance to research conventions
• Code index has 0 modules? → Run aiframework index --target .
• No learnings captured? → Start using /aif-learn after discoveries
• CLAUDE.md over 200 lines? → Run aiframework refresh to slim it down

Step 6: Generate Status Report

Output a clean table:

═══════════════════════════════════════════════════
  aiframework Status — {project_name}
═══════════════════════════════════════════════════

  Stack:      {language} / {framework}
  Archetype:  {archetype} ({maturity}, {complexity})
  Tier:       {tier}

  ┌─────────────────────────┬──────────┐
  │ Component               │ Status   │
  ├─────────────────────────┼──────────┤
  │ CLAUDE.md               │ ✓ (N lines) │
  │ AGENTS.md               │ ✓ (N lines) │
  │ .cursorrules            │ ✓ / ✗    │
  │ Extended rules          │ ✓ N files │
  │ Custom skills           │ ✓ N skills │
  │ Git hooks               │ ✓ / ✗    │
  │ CI workflow             │ ✓ / ✗    │
  │ Code index              │ N files, N symbols │
  │ Vault                   │ ✓ / ✗    │
  │ Learnings               │ N entries │
  └─────────────────────────┴──────────┘

  What to do next:
  ─────────────────

  1. /aif-enhance    ← Research your framework's conventions (run once)
  2. /aif-analyze    ← Find missing tests, circular deps (run once)
  3. Start coding!   ← aiframework is ready

  Recurring:
  4. /aif-learn "..."  ← After discovering gotchas
  5. /aif-evolve       ← Weekly: improve rules from learnings
  6. /aif-pulse        ← Weekly: discover new Claude Code features

Step 7: Offer Quick Actions

If gaps were found, offer to fix them:

• "I see your build command is NOT_CONFIGURED. Want me to detect it from your Makefile/package.json?"
• "No framework invariants found. Want me to run /aif-enhance to research your framework's best practices?"
• "CLAUDE.md is 300 lines — want me to run aiframework refresh to slim it down?"
• "No learnings captured yet. After your next bug fix, run /aif-learn to record what you learned."

Output Rules

• Be concise — status table + numbered next steps
• Don't dump raw file contents — summarize
• If everything is green, say so: "All systems go. Start coding."
• If there are gaps, prioritize: what gives the most value for least effort?