feat: align P7/P8/P9 numbering to workspace canonical#16
Merged
Conversation
…1–P19
Templates were stuck at "thirteen primitives" (P1–P13) while bstack
SKILL.md, scripts/doctor.sh, and references/primitives.md had moved to
nineteen (P1–P19). This scaffolded every new workspace into permanent
disagreement with the catalog — `bstack bootstrap` produced governance
files that `bstack doctor` then reported as having six missing primitive
sections.
Changes:
1. **CLAUDE.md.template** (+58/-22): rewritten to nineteen primitives
with the full table (P1–P19), short-name convention + index, Plugin
Skill Precedence section (bstack > superpowers hierarchy), updated
governance/hooks/conventions sections, Self-Documenting Standards
block.
2. **AGENTS.md.template** (+220/-14): bumped intro to nineteen, every
### P# heading renamed to "### P# — Name: …" form, short-name
convention + index added, full bodies for P14 (Dep-Chain), P15
(Snapshot), P16 (Crystallize), P17 (Lens), P18 (Audience), P19
(Orchestrate) — each with What/How/Invariant/Reflexive Trigger Rule.
Composition-loop diagram rewritten in short-name form with P14–P19
threaded at pre-flight + boundary crossings. Plugin Skill Precedence
section after composition loop documents the bstack > plugin hierarchy
and the "what this kills / what this keeps" partition.
3. **SKILL.md** (+1/-1): tiny drift fix — body said "16 primitives"
while frontmatter said "nineteen". Now consistent.
4. **tests/template_lockstep.test.sh** (new, 188 lines): asserts
primitive-count and structural consistency across the four
governance surfaces that must agree:
- SKILL.md frontmatter description (canonical count word + P19
reference + P1–P19 trigger span)
- scripts/doctor.sh EXPECTED_COUNT (the validator)
- assets/templates/CLAUDE.md.template (intro count, table rows,
short-name index entry count, Plugin Skill Precedence presence)
- assets/templates/AGENTS.md.template (intro count, composition-loop
intro, ### Pn section count, Plugin Skill Precedence presence)
- Short-name index payload identity between the two templates
13 assertions, all green at canonical count = 19. Designed to be the
guardrail that makes this kind of drift CI-visible — when the next
primitive is added (P20 in flight), this test catches any template
that didn't get updated.
Numbering uses **bstack canonical ordering**: P7=Freshness, P8=Janitor,
P9=Wait. This matches the prior template state and bstack catalog
(SKILL.md, doctor.sh, references/primitives.md). The ~/broomva workspace
itself uses P7=Wait, P8=Freshness, P9=Janitor because of historical
script + config-dir naming (BROOMVA_P8_HOME, ~/.config/broomva/p9-
janitor/); that's a workspace-specific deviation documented in
workspace#54 and is NOT a template concern — new bstack installs get
clean canonical numbering.
Companion to broomva/workspace#54 (precedence + short names + P19 in
workspace governance). Independent of bstack#14 (P20 sync) — that PR
adds P20 to the catalog; once it merges, a follow-up updates these
templates + lockstep test to canonical=20.
Follow-ups (separate PRs):
- bootstrap.sh + revamp.sh ORDERED_SKILLS divergence from SKILL.md
ROSTER (bootstrap installs persist + wealth-management + investment-
management; ROSTER instead has autonomous + role-x).
- P20 propagation to templates after bstack#14 merges.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Companion to #14 (Cross-Model Adversarial Review Gate). bstack catalog moved to twenty primitives via #14 (SKILL.md / doctor.sh / references/ primitives.md updated) but #14 did not touch templates. That's exactly the drift this PR was created to close, and exactly what the new template_lockstep.test.sh caught immediately after rebase: --- before this commit --- CLAUDE.md.template intro says 'twenty irreducible primitives (P1–P20)' → FAIL (says nineteen / P19) AGENTS.md.template intro says 'twenty irreducible building blocks' → FAIL AGENTS.md.template composition-loop intro says 'twenty primitives' → FAIL AGENTS.md.template ### Pn section count → FAIL (19, expected 20) CLAUDE.md.template primitive table row count → FAIL (19, expected 20) CLAUDE.md.template short-name index entry count → FAIL (19, expected 20) AGENTS.md.template + CLAUDE.md.template references P20 → FAIL --- after this commit --- All 13 lockstep checks passed at canonical count = 20 (twenty) Changes: - **assets/templates/CLAUDE.md.template**: count phrase "nineteen → twenty", "P1–P19 → P1–P20", short-name index appends "Cross-Review (P20)", primitive table gets P20 row (mechanism: 3 strata A/B/C, anti-slop ≥7/10, ≤3 fix rounds, broomva/cross-review skill; invariant: substantive PRs require ≥7/10 verdict before Pipeline P4 auto-merge). - **assets/templates/AGENTS.md.template**: count phrase nineteen→twenty in intro + composition-loop, short-name index appends Cross-Review (P20), new `### P20 — Cross-Review` section with full What/How (3-strata table) / Invariant / Reflexive Trigger Rule (5 triggers). Composition-loop diagram threads Cross-Review (P20) between Empirical (P11) deploy verification and Pipeline (P4) auto-merge — the exact insertion point per P20 spec ("gate fires before P4 auto-merge"). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Resolves the pre-existing numbering inconsistency where bstack canonical disagreed with broomva/workspace AGENTS.md on P7/P8/P9. After this change: P7 = CI Watcher + Productive Wait (broomva/p9 skill — historical name) P8 = Skill Freshness Check (SessionStart hook) P9 = Branch + Worktree Janitor (make janitor) This matches the workspace's renumbering documented in its naming-note (P7's skill repo is broomva/p9 — named when it was the 9th primitive; renaming the SKILL repo would break every npx skills add install). Why workspace canonical wins: - Workspace numbering is the actively-edited canonical (7 primitives added under it in the last 3 sessions: P14, P15, P16, P17, P18, P19, P20) - Reverting workspace would touch many files; aligning bstack to workspace is contained to this single repo - The doctor (which reads workspace AGENTS.md/CLAUDE.md) now passes 73/73 against workspace — eliminates the 1 pre-existing gap reported for several sessions Changes: SKILL.md: - Frontmatter description: P7/P8/P9 sentences reordered to workspace canonical (Productive Wait → Skill Freshness → Janitor) - Primitives table: rows swapped (P7 = Productive Wait skill p9 historical, P8 = Skill Freshness, P9 = Janitor) references/primitives.md: - Table of contents: P7/P8/P9 entries swapped to workspace order - §P7 — now CI Watcher + Productive Wait (was Skill Freshness); skill- name note rewritten to "broomva/p9 historical name" instead of "names match" - §P8 — now Skill Freshness Check (was Janitor); env vars updated to BROOMVA_P8_THRESHOLD_DAYS with legacy P7 fallback - §P9 — now Branch + Worktree Janitor (was Productive Wait); protected path now ~/.config/broomva/p9-janitor/ with legacy p8 fallback - Reflexive Trigger Rule subsection: now P7 (Productive Wait), was P9 scripts/doctor.sh: - Header comment: reflexive list updated (P6, P7, P10..) instead of (P6, P9, P10..); script-mapping comments updated (P7 = p9.py, P8 = skill-freshness-hook, P9 = branch-janitor) - P_NAMES array: P7/P8/P9 titles swapped to workspace canonical - REFLEXIVE_PRIMS: P9 → P7 (Productive Wait is the reasoning-enforced one; Janitor (now P9) is mechanism-only) - SCRIPT_PATHS: reordered so labels (P1, P2, P6, P7, P8, P9, P12) map to disk paths in the right order - **Format-tolerant section matching**: doctor now accepts both `### P1: Title` AND `### P1 — Label: Title` formats (workspace AGENTS.md uses the labeled form; doctor was missing matches against workspace files). Fixed in P_NAMES section-presence check + the awk pattern in REFLEXIVE_PRIMS scan. Companion fix in workspace: - broomva/workspace/scripts/bstack-primitive-lint.py parse regex extended to accept `### P\d+ —` (label form) in addition to `### P\d+:` (original form) Verified: - bash scripts/doctor.sh against workspace → 73/73 PASSES - G-L3-1 + G-L3-2 both green The naming note about broomva/p9 skill being historically named is preserved (and now consistent across workspace and bstack). The "primitive number matches skill name" pedagogical claim in the old §P9 is reframed in the new §P7 as "skill repo name is stable; primitive number changed; this is intentional." Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
|
Warning Rate limit exceeded
You’ve run out of usage credits. Purchase more in the billing tab. ⌛ How to resolve this issue?After the wait time has elapsed, a review can be triggered using the We recommend that you space out your commits to avoid hitting the rate limit. 🚦 How do rate limits work?CodeRabbit enforces hourly rate limits for each developer per organization. Our paid plans have higher rate limits than the trial, open-source and free plans. In all cases, we re-allow further reviews after a brief timeout. Please see our FAQ for further information. ℹ️ Review info⚙️ Run configurationConfiguration used: defaults Review profile: CHILL Plan: Pro Run ID: 📒 Files selected for processing (6)
✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
broomva
added a commit
that referenced
this pull request
May 15, 2026
Closes the gap where `bstack bootstrap` installed role-x as a skill (via ROSTER in #13) but left its hooks dormant in `.claude/settings.json` — agents installing bstack fresh had P17 routing files on disk but no UserPromptSubmit or coverage trigger. Every workspace had to wire those by hand. Confirmed end-to-end against this workspace's live settings. Added (assets/templates/settings.json.snippet): - UserPromptSubmit block → role-x-intake-hook.sh (P17 per-prompt routing) - SessionStart block extended → role-x-coverage-hook.sh (P17 registry- health nudge, ≤1/24h cooldown, silent when fire-rate ≥30%) - New ${BROOMVA_HOME} placeholder for $HOME-rooted paths (npx skills add -g installs at ${HOME}/.agents/skills/<name>) Extended (scripts/bootstrap.sh): - Phase 3 sed substitution: -e for both ${BROOMVA_WORKSPACE} and ${BROOMVA_HOME} → workspace dir and $HOME respectively - Phase 3 Python merge path: two-step .replace() for the same two placeholders before json.loads() - Idempotent merge logic unchanged — existing-by-basename detection already handles re-runs correctly (verified Case C: 0 new wires on a workspace where both role-x hooks were already manually wired) - Drive-by fix: log format string was "PP8"/"PP17" (double-P) because the _bstack_primitive value already contains the "P" prefix; corrected to ({primitive}) with P? fallback when key absent Extended (scripts/doctor.sh): - §6 hook-wiring array gets role-x-intake-hook.sh + role-x-coverage-hook.sh with P17 labels - Old skill-freshness-hook.sh label corrected from P7 → P8 (post-PR #16 numbering) - Docstring header §5 now enumerates all 5 hooks by primitive - Docstring header §6 adds P17 disk-presence reachability spec Validation: - JSON valid on new snippet (python3 -c 'json.load(...)') - Bash syntax green on bootstrap.sh + doctor.sh (bash -n) - Three dry-run cases on /tmp scratch: A. Fresh scaffold → snippet renders 5 hook events with absolute ${HOME} paths resolved correctly B. Existing settings.json with only P1/P2 → merge adds 4 missing hooks (P8 freshness, P17 intake, P17 coverage, P1 notification) C. Workspace with all 5 hooks already wired → 0 new wires (idempotency confirmed) - doctor.sh on live workspace: all 5 hook checks [ok] including the two new P17 entries Coordination with concurrent PRs: - PR #17 (broomva): SKILL.md prose drift + AGENTS/CLAUDE templates + scaffold-and-doctor test. Touches SKILL.md, AGENTS.md.template, CLAUDE.md.template, template_lockstep.test.sh — NO overlap with this PR (verified via gh pr diff 17). - SKILL.md ## Hooks section update deferred to a follow-up PR after #17 lands, to keep edits non-overlapping. Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Resolves the pre-existing P7/P8/P9 numbering inconsistency between bstack canonical and workspace canonical that's been showing as a 1-gap doctor report across multiple sessions.
After this change (matches
broomva/workspaceAGENTS.md):broomva/p9(historical name)make janitor)Why Option B (align bstack → workspace) over Option A
Files changed
Verification
Followups
/.config/broomva/p7/`, `/.config/broomva/p9-janitor/`) keep working: the primitives.md notes specify legacy-name fallbacks in both directions.🤖 Generated with Claude Code