feat(conductors): Mind + Cortex dashboards emit the state.json cockpit feed (#418)

[Jammy2211]

Summary

Organ cockpit (PyAutoBrain#418, PR C of 3 — merge last, after PyAutoMind and PyAutoCortex): the intake and cortex conductors render the state.json v1 cockpit feed beside the Mind and Cortex dashboards they already write, using board/_state.build_state. dashboard --apply writes it, dashboard --check covers it (with updated normalised out, like the generation stamp), cortex checkin carries it. Never re-derives: items reuse the rows' own copy payloads.

Also fixes a real regression found on the way: _mind_home expected github: on the line after PyAutoMind: in repos.yaml, but since the grouped-checkouts layout (Mind 69174a65, 2026-09-19) path: comes first, so the Mind home resolved to "" and the live dashboard silently lost its Pages pointer and absolute links. The regex now scans the whole indented entry. Expect the Mind self-heal after this merge to commit a large dashboard.html link diff along with state.json.

API Changes

• agents/conductors/intake/_intake.py: render_state(c) (+ _state_status, _state_items, _start_here_picks, _now_iso clock seam, _state_body/_page_body); pages gains state.json; --json includes state. Status: red on registry drift or any in-flight row awaiting-input/blocked; yellow on awaiting-merge/library-shipped rows, human-review rows, pending-release rows or picks; grey on an empty Mind; else green. Headline N picks · M in flight.
• agents/conductors/cortex/_cortex.py: render_state(c) (+ checkin_stale — Python port of the page's 180-min JS rule, _problem_ledger, _state_items, _now, state_body/page_body); render_pages adds state.json. Status: red on problems; yellow on running runs or a stale check-in; grey with no ledgers; else green. Headline A active · R running.
• _mind_home regex fix (above).
  See full details below.

Test Plan

• pytest tests/test_intake_dashboard.py tests/test_cortex_conductor.py tests/test_state_feed.py -q — 196 passed (5 + 6 new, 2 new fixtures)
• Full suite 1037 passed; 1 env-only failure (test_hygiene_map_and_heart_board_find_grouped_checkouts) with activate.sh's PYAUTO_MIND exported, passes unset — pre-existing
• Against the worktree Mind: --apply dashboard → "Wrote: dashboard.md + dashboard.html + state.json (237 prompts)"; --check → current; validator state: ok (yellow — 24 picks · 7 in flight, 20 items)
• Against the worktree Cortex: --apply → 12 ledgers, 7 running; --check → current; validator state: ok (yellow — 7 active · 7 running, 8 items)
• Tenant firewall OK
• After merge: Mind + Cortex dashboard_refresh.yml commit state.json and dispatch Pages; both pages_dashboard.yml green with state: ok; both published feeds validate

Full API Changes (for automation & release notes)

Added

• intake: render_state, _state_status, _state_items, _start_here_picks, _now_iso, _state_body, _page_body; state.json in pages (check/apply/json)
• cortex: render_state, checkin_stale, _problem_ledger, _state_items, _now, state_body, page_body; state.json in render_pages
• tests/fixtures/state/{mind,cortex}.json; docs lines in both conductors' AGENTS.md and board/AGENTS.md

Changed

• _mind_home: scans the whole PyAutoMind: entry for github:

Removed

• none

Migration

• none — additive; dashboard.md/html rendering unchanged apart from the restored absolute links

Heart RED override (development only)

Heart readiness is RED (snapshot 2026-09-26T11:27:07+00:00). Verbatim reason string:

• release validation FAILED (stage integrate)

Pre-existing release-integrate failure, unrelated to this diff. Branch gates passed as listed above.

Human authorization (live, this session, 2026-09-26, for PyAutoBrain#418): "override and continue" — development shipping + merge on green checks (Mind → Cortex → Brain); no release

Generated by the PyAutoLabs agent workflow.

🤖 Generated with Claude Code