docs: Jigo punch list — close 3 majors from doc-canon audit #3

Merged

j7 merged 3 commits from docs/jigo-punch-list-3-fixes into main

2026-07-09 15:25:36 -04:00

san commented

2026-07-09 04:47:52 -04:00

Collaborator

Summary

Applies the 3 majors from Jigo's review of PR-1/PR-2 (kanban card t_377762_b187).

Verdict from Jigo: Ready with minor fixes.

What changed

kanban/backlog.md: 12 stale [ ] checkboxes flipped to [x] (E0-T1..T5, E1-T1..T7) — all closed by PR-1/PR-2 commits. Adds E1-T8 (Jigo review row) closed by this PR.
docs/external-workers.md: §2 vs §5 contradiction fixed. §2 clarifies owner_wolf is the named responsible, while Okkoto is the fixed operator per §5.

Diff stat

 docs/external-workers.md |  4 ++--
 kanban/backlog.md        | 25 +++++++++++++------------
 2 files changed, 15 insertions(+), 14 deletions(-)

Major 1 downgraded (no change)

The 3 references to prompts/jigo-review.md (in docs/playbook.md, docs/workflow.md, prompts/san-handoff.md) are all valid — file exists at the path (313 bytes). Thin but not broken. Substantive review-lens doc is a Phase E1 follow-up.

Out of scope (4 minors deferred)

Per Jigo's framing ("what can wait until after Sprint 0"):

GBrain card dedup between kanban/backlog.md E2-T1 and specs/gbrain-substrate.md
profile.yaml canonical-location clarification (~/.hermes/wolf-souls/_profile-template.yaml)
Honcho row wording in specs/starros-stack-spec.md §3 vs decisions/0003 glossary
Mission Control source-dir pointer (hermes-workspace-src)

The 400-char dispatcher truncation of Jigo's audit summary makes in-line reconstruction of the minor why-it-matters / fix lines speculative, so they are deferred rather than fabricated.

Refs

Kanban card: t_377762_b187 on starros-v1 board
Matrix status: !JLSvRbGBirIsCCXjOb:mtrx.starrwulfe.xyz
Base: PR-2 tip b55107d

Per J7 directive, no reviewers assigned yet.

## Summary Applies the 3 majors from Jigo's review of PR-1/PR-2 (kanban card `t_377762_b187`). **Verdict from Jigo**: Ready with minor fixes. ### What changed - **`kanban/backlog.md`**: 12 stale `[ ]` checkboxes flipped to `[x]` (E0-T1..T5, E1-T1..T7) — all closed by PR-1/PR-2 commits. Adds E1-T8 (Jigo review row) closed by this PR. - **`docs/external-workers.md`**: §2 vs §5 contradiction fixed. §2 clarifies `owner_wolf` is the *named responsible*, while Okkoto is the *fixed operator* per §5. ### Diff stat ``` docs/external-workers.md | 4 ++-- kanban/backlog.md | 25 +++++++++++++------------ 2 files changed, 15 insertions(+), 14 deletions(-) ``` ### Major 1 downgraded (no change) The 3 references to `prompts/jigo-review.md` (in `docs/playbook.md`, `docs/workflow.md`, `prompts/san-handoff.md`) are all valid — file exists at the path (313 bytes). Thin but not broken. Substantive review-lens doc is a Phase E1 follow-up. ### Out of scope (4 minors deferred) Per Jigo's framing ("what can wait until after Sprint 0"): 1. GBrain card dedup between `kanban/backlog.md` E2-T1 and `specs/gbrain-substrate.md` 2. `profile.yaml` canonical-location clarification (`~/.hermes/wolf-souls/_profile-template.yaml`) 3. Honcho row wording in `specs/starros-stack-spec.md` §3 vs `decisions/0003` glossary 4. Mission Control source-dir pointer (`hermes-workspace-src`) The 400-char dispatcher truncation of Jigo's audit summary makes in-line reconstruction of the minor *why-it-matters* / *fix* lines speculative, so they are deferred rather than fabricated. ### Refs - Kanban card: `t_377762_b187` on `starros-v1` board - Matrix status: `!JLSvRbGBirIsCCXjOb:mtrx.starrwulfe.xyz` - Base: PR-2 tip `b55107d` Per J7 directive, no reviewers assigned yet.

san added 3 commits

2026-07-09 04:47:52 -04:00

docs: split canonical spec from runtime state, add 5 companion docs 8e376e5bfa

v2.0 of specs/starros-stack-spec.md (14088 bytes, was 26995):
- Adds §0 'Scope: target architecture vs current implementation state'
  so the spec stops absorbing debug-notebook cross-references.
- Trims runtime facts (port maps, broken services, drift) out of
  architecture — moves them to docs/runtime-state.md.
- Renames §14 to 'Recommendations' (3 spec-shape only) and drops the
  Ashitaka debug-action recommendation (it was a runtime concern, not
  architectural).
- Adds ADR 0003 link in §12.

5 new companion docs:
- docs/runtime-state.md (live operational state; per-week rewrite)
- specs/gbrain-substrate.md (project substrate, in-scope per J7 2026-07-09)
- docs/persistence-boundaries.md ('where do I write this?' map)
- docs/external-workers.md (CLI workers policy; 5 in scope)
- decisions/0003-canonical-spec-boundaries.md (rationale + glossary)

Cross-references in the spec point at the companion docs; spec is now
the target architecture only.

Refs: ADR 0001 (repo is source of truth), PR #1 (commit 21dee6d),
specs/starros-stack-spec.md v1.1.

docs: cross-file normalize — pointers to companion docs, E1 epic, packet field, ADR-0003 link b55107d86b

- kanban/backlog.md: drop E1 (Mission Control — paused), add E1
  doc-canon epic (T1-T7), rename E2-T1 OKF/GBrain → GBrain, add E3
  runtime-ops epic mapped to phases P0-P13.
- packets/leaf-card-template.md: add 'Report-back channel' field
  (default: project's Matrix room; see docs/external-workers.md §6).
  Honor docs/external-workers.md in Constraints.
- prompts/san-handoff.md: expand reading list to 15 docs including the
  5 new companion docs (runtime-state, persistence-boundaries,
  external-workers, gbrain-substrate, ADR 0003).
- prompts/fable-global.md: add 'Paused components' callout with
  cross-reference to ADR 0003 §Glossary.
- docs/playbook.md: add 'Companion docs' section pointing at the 5
  new docs; rename 'Status: Draft v0' → 'v1'; add 'canonical' line
  at top.
- docs/workflow.md: tighten to canonical doc references; rename
  'Claude Code' → 'external coder workers' per docs/external-workers.md
  §7 naming normalization.
- decisions/0001-repo-is-source-of-truth.md: add 'Related: ADR 0003'
  and stronger consequences (per-week doc-canon pattern).

docs: Jigo punch list (PR-3) — flip closed doc-canon checkboxes, clarify external-workers operator authority 28527c758a

Applies the 3 majors from Jigo's review of PR-1/PR-2 (kanban card t_377762_b187):

1. [major] Stale backlog checkboxes: E0-T1..T5 and E1-T1..T7 are all closed by
   PR-1/PR-2 commits but still showed [ ]. Now [x]. Adds E1-T8 (Jigo review row)
   closed by this PR.

2. [major] docs/external-workers.md §2 vs §5 contradiction: §2 said the owner_wolf
   is the operator; §5 said Okkoto is the operator. These are not equivalent.
   Fix: §2 clarifies owner_wolf is the *named responsible*, with Okkoto fixed as
   the actual operator per §5. §5 calls out the cross-reference explicitly.

3. [major] Downgraded —  references in playbook/workflow/
   san-handoff are valid (file exists at the path, 313 bytes). Thin but not
   broken. Left as-is; substantive review-lens doc is a Phase E1 follow-up.

Out of scope (4 minors deferred per J7 'what can wait until after Sprint 0'):
- GBrain card dedup between backlog E2-T1 and specs/gbrain-substrate.md
- profile.yaml canonical-location clarification
- Honcho row wording in specs/starros-stack-spec.md §3 vs decisions/0003
- Mission Control source-dir pointer (hermes-workspace-src)

Refs: kanban card t_377762_b187, Jigo verdict 'Ready with minor fixes'.