Review Cross — Development agent for Claude Code
Cross-model correctness review of a diff, on a different model from the main session (roles.toml review-cross; Codex by default, through bin/agent-run).
How to install Review Cross
Installs to ~/.claude/agents/mshoaibmanz-agent-kit-review-cross.md
mkdir -p ~/.claude/agents && curl -fsSL https://raw.githubusercontent.com/mshoaibmanz/agent-kit/HEAD/agents/review-cross.md -o ~/.claude/agents/mshoaibmanz-agent-kit-review-cross.md Restart Claude Code, or start a new session, for it to be picked up.
What Review Cross does
name: review-cross description: Cross-model correctness review of a diff, on a different model from the main session (roles.toml review-cross; Codex by default, through bin/agent-run). Pass the objective and the diff scope. Read-only; finds the strongest reasons the change should not ship. Finding ids CX-. tools: Read, Grep, Glob, Bash
You are the cross-model reviewer, the second reviewer of a code change. Another review agent, on a different model, may audit the same change inAlternatives in Development
- SLO Implement — You are an SLO (Service Level Objective) expert specializing in implementing reliability standards a... - wsho 31.9k ★
- Civitai Correctness Review — Reviews a feature segment in the main Civitai Next.js app (src/) for safety gaps — authorization scoping, mone 7.2k ★
- Ir Correctness Reviewer — Reviews Slang IR pass changes for correctness, SSA invariants, and type system integrity 5.6k ★
Full documentation available on GitHub
View Source RepositoryRelated Agents
Cover Paper
Use when a paper is about to be submitted to a journal and needs its submission package: (a) a Cover Letter, s
Bulk Reader
Cheap worker for I/O-heavy reading. Use instead of reading large files (>350 lines) or several files into the
Judge Same Vendor
Same-vendor different-model judge for the pair-programmer harness. Dispatches to the matching vendor's critiqu
Req Pl
Clarifies objective, non-goals, constraints, acceptance, and failure behavior before implementation when scope
Loop Evaluator
Evaluate artifact strictly against active loop criteria and return structured pass/fail output. Use proactivel
Role Triage
Reads a batch of job descriptions and returns a compact ranked shortlist. Use when more than about eight roles
Related Skills
Agency Doctor
Validate the disabled-by-default Agency orchestration policy and list installed Codex TOML roles without modif
Landscape Scan
Topic-parameterised external-intelligence scan + digest. Default topic is the AI/Claude Code/PKM landscape; pa
Vibe Check
Production resilience audit for AI-generated code. Catches edge cases, scale failures, and missing error handl