Quintet Multi-Model Code Review

Performs multi-perspective code reviews across ready AI provider CLIs for git diffs or file contents, consolidating findings into an actionable verdict.

Workflow

1. Target Preparation: Identify target git diff or file path.
2. Execute Review: Run multi-model review across ready providers:

QBIN="${CLAUDE_PLUGIN_ROOT:+$CLAUDE_PLUGIN_ROOT/bin/quintet}"
[ -n "$QBIN" ] && [ -x "$QBIN" ] || QBIN="$HOME/.gemini/config/plugins/quintet/bin/quintet"
[ -x "$QBIN" ] || QBIN="$HOME/.gemini/extensions/quintet/bin/quintet"
[ -x "$QBIN" ] || QBIN="$(command -v quintet 2>/dev/null || echo quintet)"
"$QBIN" review "$(git diff HEAD~1)" claude,agy,opencode

3. Output Validation & Fallback:

   • Validation: Confirm that review output was received from at least 1 ready provider.
   • Fallback: If a provider fails or times out, proceed with remaining findings and note the skipped model in the synthesis header.
   • Diff Size: If diff exceeds context limits, narrow scope: "$QBIN" review "$(git diff HEAD~1 -- path/to/file)".

4. Structured Synthesis & Deliverable Format: Format the deliverable strictly using the following structure:

## Verdict: [Go / No-Go / Changes Requested]

### Blocking Issues (2/3 models agree)
- **[Severity: High]** Description of critical bug, security flaw, or breaking change.

### Worth Fixing
- **[Severity: Medium]** Refactoring recommendation or performance optimization.

### Recommendation
- Summary recommendation with explicit technical reasoning.

Reference Materials

• Complete CLI subcommands, options, and setup: references/CLI_REFERENCE.md