feat: sync P20 Cross-Model Adversarial Review Gate to bstack

[broomva]

Companion to broomva/workspace#55. Brings bstack consistent: count 19→20, P20 row in primitives table, ROSTER adds cross-review (29→30 skills), doctor.sh updates, §P20 section in primitives.md with the 3-strata decision matrix.

🤖 Generated with Claude Code

Summary by CodeRabbit

• Documentation
  • Introduced a new P20 primitive enabling cross-model adversarial review with multiple review strata and anti-slop scoring thresholds
  • Updated documentation to reflect expanded primitive set (P1-P20 with 30 curated skills)
  • Updated compliance verification tooling to include the new primitive

[coderabbitai]

📝 Walkthrough

Walkthrough

This PR introduces a new primitive P20 (cross-model adversarial review gate) to the Broomva Stack, adding one primitive and one skill to the framework. The changes span the specification document, the canonical skill manifest, and the compliance validation script, updating all references to primitive enumeration to include P20 alongside P1–P19.

Changes

P20 Cross-Model Adversarial Review Gate Addition

Layer / File(s)	Summary
P20 primitive contract specification references/primitives.md	Introduces P20 as a mandatory cross-model review gate requiring three escalating review strata, anti-slop scoring with up to three fix/rescore rounds, merge gating for substantive PRs, and a reflexive trigger rule that forbids relying solely on the writing model.
SKILL.md manifest and primitives table updates SKILL.md	Updates metadata to reflect 20 primitives (P1–P20) and 30 skills, extends contract narrative and trigger conditions to reference P1 through P20, adds P20 entry to the primitives table documenting the cross-review behavior, and adds cross-review to the onboarding roster check.
Doctor script compliance validation updates scripts/doctor.sh	Expands script header to cover P1–P20, updates CLAUDE.md primitives table validation from 19 to 20, adds P20 section to AGENTS.md primitive enumeration, and includes P20 in reflexive trigger enforcement checks.

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~20 minutes

Possibly related PRs

• broomva/bstack#11: Updates the same primitive enumeration plumbing (SKILL.md, references/primitives.md, scripts/doctor.sh) to add P17/P18, following the same structural pattern as this P20 addition.
• broomva/bstack#5: Extends primitive-enumeration plumbing (scripts/doctor.sh and SKILL.md) to add P12, sharing the same compliance-validation pattern as this PR's P20 integration.
• broomva/bstack#9: Updates SKILL.md primitives table and scripts/doctor.sh reflexive-trigger enforcement to add P14–P16, using the same integration approach for new primitives.

Poem

🐰 A new gate arrives, P20 bright,
Cross-model guardians keeping slop in sight,
Three strata strong, no self-echo's plight,
The stack grows true, from twenty heights of might! ✨

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The pull request title accurately summarizes the primary change: adding P20 (Cross-Model Adversarial Review Gate) to the bstack Stack contract and synchronizing related documentation and checks.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch feat/p20-cross-model-review

---

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.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

Actionable comments posted: 3

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@references/primitives.md`:
- Line 388: Add a TOC entry linking to the new section titled "P20 — Cross-Model
Adversarial Review Gate": locate the table of contents in
references/primitives.md and insert a markdown link that points to the section's
anchor text matching the exact heading "P20 — Cross-Model Adversarial Review
Gate" so the TOC item navigates to that heading.
- Around line 400-407: The P20 gate text is self-contradictory because the
header requires "verdict logged in PR comments + Linear ticket" while the P20
Reflexive Trigger Rule step 1 ("Before pushing substantive PRs — fire the
gate... Score + verdict precede push") mandates a pre-push verdict; resolve by
explicitly splitting responsibilities: require a mandatory pre-push local
score+verdict (retain the language in "Before pushing substantive PRs" and the
phrase "Score + verdict precede push") and separately require post-push PR
comment + Linear ticket logging of that pre-push verdict (or, alternatively,
move the gate to be a mandatory pre-merge check and update the header to remove
"PR-comment logging" as a prerequisite); update the P20 heading and step 1 to
mention "pre-push local verdict" and add a new step (or modify existing logging
clause) that records that verdict in PR comments + Linear ticket after push.

In `@SKILL.md`:
- Line 52: Update all occurrences in SKILL.md that reference the primitive and
skill counts so they match the new canonical wording "Twenty irreducible
primitives. Thirty curated skills." Search for variants like "Twenty / Thirty",
older numeric values, or phrases mentioning primitives/skills elsewhere in the
document (e.g., the headings and descriptive sentences that currently disagree)
and replace them with the exact canonical phrase; ensure any pluralization or
surrounding context remains grammatically correct and that cross-references or
summaries (such as counts in lists, headers, and the one-line tagline) are
normalized to the same numbers.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: fe1eaeb4-ddc7-4a3f-a078-8325151838ce

📥 Commits

Reviewing files that changed from the base of the PR and between f412105 and 84c5713.

📒 Files selected for processing (3)

• SKILL.md
• references/primitives.md
• scripts/doctor.sh

[coderabbitai]

⚠️ Potential issue | 🟡 Minor | ⚡ Quick win

Add P20 to the table of contents for discoverability.

The new section is present, but the TOC does not link to it.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@references/primitives.md` at line 388, Add a TOC entry linking to the new
section titled "P20 — Cross-Model Adversarial Review Gate": locate the table of
contents in references/primitives.md and insert a markdown link that points to
the section's anchor text matching the exact heading "P20 — Cross-Model
Adversarial Review Gate" so the TOC item navigates to that heading.

[coderabbitai]

⚠️ Potential issue | 🟠 Major | ⚡ Quick win

Resolve P20 gate timing contradiction (pre-push vs PR-comment logging).

Line 400 requires PR-comment logging, but Line 406 requires verdict before push. Those cannot both be true without a “local pre-push score + post-push PR logging” split, or by moving the mandatory gate to pre-merge.

Proposed wording adjustment

- Scoring: anti-slop ≥ 7/10 to pass; max 3 fix rounds; verdict logged in PR comments + Linear ticket. Implementation: `broomva/cross-review` skill.
+ Scoring: anti-slop ≥ 7/10 to pass; max 3 fix rounds; verdict logged in Linear ticket and, once a PR exists, mirrored in PR comments. Implementation: `broomva/cross-review` skill.

- 1. **Before pushing substantive PRs** — fire the gate (Strata A if Codex, else B+C). Score + verdict precede push.
+ 1. **Before merging substantive PRs** — fire the gate (Strata A if Codex, else B+C). Score must pass before merge; log verdict to Linear and PR comments.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@references/primitives.md` around lines 400 - 407, The P20 gate text is
self-contradictory because the header requires "verdict logged in PR comments +
Linear ticket" while the P20 Reflexive Trigger Rule step 1 ("Before pushing
substantive PRs — fire the gate... Score + verdict precede push") mandates a
pre-push verdict; resolve by explicitly splitting responsibilities: require a
mandatory pre-push local score+verdict (retain the language in "Before pushing
substantive PRs" and the phrase "Score + verdict precede push") and separately
require post-push PR comment + Linear ticket logging of that pre-push verdict
(or, alternatively, move the gate to be a mandatory pre-merge check and update
the header to remove "PR-comment logging" as a prerequisite); update the P20
heading and step 1 to mention "pre-push local verdict" and add a new step (or
modify existing logging clause) that records that verdict in PR comments +
Linear ticket after push.

[coderabbitai]

⚠️ Potential issue | 🟠 Major | ⚡ Quick win

Normalize stale primitive/skill counts across the rest of this document.

This update correctly sets “Twenty / Thirty,” but multiple later sections still say older values (e.g., Lines 60, 61, 74, 84, 246, 295, 302, 312, 319, 393, 395). That leaves conflicting operator guidance in one canonical file.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@SKILL.md` at line 52, Update all occurrences in SKILL.md that reference the
primitive and skill counts so they match the new canonical wording "Twenty
irreducible primitives. Thirty curated skills." Search for variants like "Twenty
/ Thirty", older numeric values, or phrases mentioning primitives/skills
elsewhere in the document (e.g., the headings and descriptive sentences that
currently disagree) and replace them with the exact canonical phrase; ensure any
pluralization or surrounding context remains grammatically correct and that
cross-references or summaries (such as counts in lists, headers, and the
one-line tagline) are normalized to the same numbers.