cross-reviewlisted

bstack P20 — Cross-Model Adversarial Review Gate. The model that wrote the code cannot be the final judge of the code. Before substantive PRs merge, fire a cross-model adversarial gate — different evaluator than writer, anti-slop scoring ≥7/10, max 3 fix rounds, verdict logged in PR. Three strata: (A) Codex CLI cross-vendor for true different-model verdict, (B) fresh-context subagent under devils-advocate brief, (C) composed existing adversarial-review skills always parallel. Use cross-review when: (1) about to push a substantive PR (>200 LOC OR public API OR multi-file OR governance-class), (2) reviewing a draft plan/design before implementation, (3) auditing a feature spec against single-model blind spots, (4) integrating with the /autonomous skill's pre-push gate. Triggers on 'cross-review', 'P20', 'adversarial review', 'anti-slop', 'cross-model gate', 'different evaluator', 'devils advocate gate', 'self-review prohibition'.
broomva/skills · ★ 3 · AI & Automation · score 72

Install: claude install-skill broomva/skills

# cross-review — bstack P20 Cross-Model Adversarial Review Gate **The writer cannot be the final judge of the work.** When the same AI model plans, implements, and reviews, it will not challenge its own assumptions. It has systematic blind spots baked into its training. A different model — trained differently, with different biases and pattern preferences — catches what the first one misses. `cross-review` is the bstack gate that enforces this discipline. Substantive PRs cannot merge until a *different evaluator* than the writer scores the work ≥ 7/10 against an anti-slop rubric. ## Origin Inspired by [Dallionking/cross-model-agents](https://github.com/Dallionking/cross-model-agents) (May 2026) — 31-agent bidirectional Claude↔Codex review system. That project ships specific agents and hooks. `broomva/cross-review` absorbs the *discipline* while composing with the existing bstack adversarial-review skill toolkit. ## The 3 strata Different mechanisms for different environments. The *substance* is the gate — what mechanism implements it is secondary. | Strata | Mechanism | When | Strength | |---|---|---|---| | **A — True cross-vendor** | `codex exec -m gpt-5.4` (or similar) reads the diff and scores | Codex CLI installed | Strongest — different weights, different training, genuinely different blind spots | | **B — Cross-context same-model** | Fresh `Agent` subagent under devil's-advocate brief reads diff and scores | Always available | Weaker than (A) but still strong —