Bee Review
Run heterogeneous multi-model code review (Claude, OpenCode, Codex) with role split and severity aggregation before merge or release.
Overview
bee-review is a plugin marketplace for the Ship phase that runs cross-vendor multi-model code review with role splits and severity-tiered findings.
What is this marketplace?
- Cross-vendor swarm review combining Claude, OpenCode, and Codex in one workflow
- Role-based division of review labor across models
- Severity-tiered aggregation to prioritize fixes over noise
- Positioned to reduce missed issues versus single-model review
- Ship-phase plugin v2.0.0 from focus-tech bee-review repository
- Marketplace ships 1 plugin: bee-review at version 2.0.0
- Designed for three model families: Claude, OpenCode, and Codex
What problem does it solve?
One-model code review misses nuanced bugs and floods you with low-severity noise when you are trying to merge solo.
Who is it for?
Builders who already use Claude Code and want a structured multi-model review swarm before merge or deploy.
Skip if: Projects that only need lint autofix with no LLM review budget or no second model provider configured.
What do I get? / Deliverables
After installing bee-review, you get aggregated, severity-ranked review output from Claude, OpenCode, and Codex-style roles in a single coordinated pass.
- Consolidated review report with severity-layered findings
- Role-attributed feedback from multiple model backends
Plugins in this marketplace
1 plugin — install individually after you add the marketplace.
Recommended Marketplaces
Journey fit
bee-review is a ship-time quality gate focused on finding defects and risk in code changes, which maps cleanly to the Ship phase even if you rerun reviews during operations. The plugin description centers on structured Review with severity layering, not unit tests or launch checklists, so review is the canonical subphase.
How it compares
Multi-model review orchestration plugin, not a static linter skill.
Common Questions / FAQ
Who is Bee Review for?
It is for developers and small teams shipping from Claude Code who want parallel review from more than one AI model vendor with unified severity output.
When should I use Bee Review?
Use it before merging feature branches, after large agent-generated diffs, or when a single-model review keeps missing security or logic issues.
How do I add Bee Review to my agent?
Add the heresun/SuperReview marketplace, install the Bee Review plugin (v2.0.0) from focus-tech/Bee Review, configure access for Claude, OpenCode, and Codex as required, then run review on your target branch or PR.