Status

Status — Where Am I?

The pipeline dashboard. Reads on-disk state and reports. Never writes approval records — a deterministic filesystem read, no generative step. Runs any time.

Inputs (filesystem only)

• .genesis/project-state.md (active feature, mode)
• .genesis/features/<feature>/approvals/master.json (feature-level phases)
• .genesis/features/<feature>/approvals/group-{N}.json (group progress)
• micro-specs/*.md frontmatter (per-spec status)
• units.json (lowfield unit progress)
• git status / git log (uncommitted work, recent commits)
• gh for pending PRs (if available); a NEEDS_CLARIFICATION marker scan

Process

1. Read project-state.md for the active feature and mode.
2. Read master.json; count micro specs per group from the micro-specs/ directory.
3. If in implementation, read the active group-{N}.json for spec-level detail.
4. If lowfield, read units.json for unit progress.
5. Check for pending PRs and NEEDS_CLARIFICATION markers.
6. Render the dashboard (or JSON with --json). --all shows every feature.

Output

FEATURE: <name>   BRANCH: ...   MODE: <mode>   PHASE: ...
Pipeline:
  [x] Brainstorm / Design / Specify / North Star / Micro Specs
  [ ] Implement — G0: 4/4, G1: 2/6  <- current
  [ ] Audits — G0: A1+ A2+ A3+ A4+ A5+, G1: pending
  [ ] Final (F-pass)
Progress: 6/26 micro specs implemented   PRs: 2 pending   Blockers: none
Approval State: | Phase | Status | Disposition | Date | Notes |

Mode differences

Brownfield adds scope boundaries (files in/out of scope) and regression surface. Lowfield adds a unit-progress table with per-unit stage depth (e.g. 6/8 line-edit).

Natural language triggers

- "status"
- "where am I"
- "pipeline position"