[Jammy2211]

Closes #553. Phase 2 of the autonomous-orchestrators epic: the organism now parses what phase 1 wrote. Read with PyAutoLabs/PyAutoMind#523 (merge that one first).

What changes

• Sizing (_sizing.py): LANE_VALUES = ("any","local-dev","ral","gpu","wsl-data"), lane_is_local(); first valid declaration wins; an invalid value (the archived any | local-dev) is recorded as lane_invalid and read as any.
• Batch planner (_batch.py): _lane_ok routes through lane_is_local; the BatchDecision carries and prints lane_counts ("Ready by lane"); the other-lane line names resources (3 laptop-lane task(s) are ready (2 ral, 1 gpu); other_lane_ready stays an int for existing readers).
• batch brief --check <file> (new _brief.py): every error names its key — review-at/dispatched ISO and ordered, goals, backlog: none|fill-idle|scoped (+scoped:), integer review-minutes/merge-cap/token-budget, lanes ⊆ lane values (cloud refused), ral-allowance with ral, the calibration rule (previous record's review-minutes-actual:), and merge-cap > 0 refused when every named task is judge and backlog: none (the human-approved refinement — a dip can still pick glance work). Unresolved named tasks warn, never refuse. Exit 0 / 1.
• batch plan --brief <file>: refuses a brief that fails the check; named tasks outrank the queue, the dip admits only what the policy covers (excluded by the brief / outside the brief's scoped backlog / lane … not in the brief's lanes), the brief's review-minutes is the budget.
• Intake (_intake.py): declared Filed:/Epic:/Phase:/Blocked-by:/Closes-when: pass through after Status:, Lane: after Unattended:; lane inference RAL/SLURM/sbatch/squeue/euclid_jump → ral, CUDA/laptop GPU → gpu, /mnt/c/Workspace/Science → wsl-data (declared wins; any not written). This is the header half of draft/bug/pyautobrain/intake_classify_file_drops_declared_epic_phase.md; its Memory:-gating half stays open and is re-scoped at close-out.
• Census: by_lane (invalid lanes counted as any + a hygiene flag), By lane: in intake census, a "Lanes:" line on the dashboard Backlog (md + html). The Mind dashboard re-renders with it on the next refresh.
• Docs: batch skill "Brief verbs" (dispatch/digest still marked planned), conductor lane-detection note.
• Tests: tests/test_orchestrators_lane_brief.py, 28 tests across witness clauses (a)–(e).

Tier: glance — merge mode: auto-merge on green if the Witness passes (human-approved plan 2026-11-03; Heart YELLOW reason set acknowledged by the human at ship).

Witness

$ pytest tests/test_orchestrators_lane_brief.py -v  (witness clauses a–e)
test_every_lane_value_parses[any]  PASSED
test_every_lane_value_parses[local-dev]  PASSED
test_every_lane_value_parses[ral]  PASSED
test_every_lane_value_parses[gpu]  PASSED
test_every_lane_value_parses[wsl-data]  PASSED
test_archived_two_value_spelling_is_rejected_and_recorded  PASSED
test_first_valid_declaration_wins  PASSED
test_resource_lanes_are_laptop_only[ral]  PASSED
test_resource_lanes_are_laptop_only[gpu]  PASSED
test_resource_lanes_are_laptop_only[wsl-data]  PASSED
test_resource_lanes_are_laptop_only[local-dev]  PASSED
test_any_runs_anywhere  PASSED
test_plan_names_the_other_lane_resources  PASSED
test_example_brief_is_accepted  PASSED
test_mind_example_validates  PASSED
test_brief_without_review_at_is_refused  PASSED
test_cloud_is_not_a_lane  PASSED
test_merge_cap_refused_when_every_named_task_is_judge  PASSED
test_ral_lane_needs_an_allowance  PASSED
test_calibration_refuses_a_brief_after_an_unmeasured_record  PASSED
test_cli_exit_codes  PASSED
test_plan_under_a_brief_puts_named_tasks_first_and_never_dips_outside  PASSED
test_intake_proposes_ral_from_the_text  PASSED
test_intake_proposes_gpu_and_wsl_data[Benchmark the CUDA path on the laptop GPU for @PyAutoArray.-gpu]  PASSED
test_intake_proposes_gpu_and_wsl_data[Read the dataset under /mnt/c/Users/x for @PyAutoArray.-wsl-data]  PASSED
test_intake_writes_no_lane_for_cloud_runnable_text  PASSED
test_declared_optional_headers_survive_apply_unchanged  PASSED
test_census_lane_histogram_totals_the_prompt_count  PASSED

$ pyauto-brain batch brief --check PyAutoMind/batches/examples/shift_brief_example.md
== Shift brief: /home/jammy/Code/PyAutoLabs-wt/autonomous-orchestrators-p2/PyAutoMind/batches/examples/shift_brief_example.md — OK ==
  ! brief.tasks: draft/feature/autolens/example_named_task.md resolves to no prompt in draft/ or active/
exit 0

$ pyauto-brain intake census | grep "By lane"   (live Mind: 289 prompts)
By lane:         any 281 · local-dev 8

full suites (PYAUTO_* unset): PyAutoBrain 1437 passed in 547.15s (0:09:07)  ·  PyAutoMind 712 passed in 252.80s (0:04:12)
repos_sync.py --check --root ~/Code/PyAutoLabs: exit 0 · lifecycle.py check: DRIFT only the pre-existing active/chapter2-explanations.md

🤖 Generated with Claude Code

[Jammy2211]

First CI run (41654587113) was red on both legs at the Tenant firewall step only (tests, discovery and line-count steps green): the new _brief.py comment named PyAutoFit#1554 and the new test file used @PyAutoArray, PyAutoFit#1/#2 and a PyAutoLabs URL. Fixed in f3fc8fc by removing those instance facts (no allowlist growth); repos_sync.py --check --only "tenant firewall (organ code)" --root <worktree> → OK, the 28 witness tests still pass. My earlier local repos_sync run used --root ~/Code/PyAutoLabs, which checked the canonical Brain rather than this branch — that is why it missed it.