Skip to content

docs: open-source AI code review tools research #270

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
Svtter merged 1 commit into from
Jun 24, 2026
+78 −0
Merged

docs: open-source AI code review tools research #270

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
78 changes: 78 additions & 0 deletions docs/alternative-review-tools-research.md
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,78 @@
# Open-Source AI Code Review Tools Research
Comment thread
Svtter marked this conversation as resolved.

> Research conducted 2026-06-24 to evaluate whether existing open-source projects can replace opencode-actions' multi-agent parallel review capability.

## TL;DR

No open-source project matches opencode-actions' multi-agent parallel review architecture. The unique value — multiple specialized reviewers (quality, security, performance, architecture, etc.) running concurrently with a coordinator synthesizing findings — has no equivalent in the open-source ecosystem.

@Svitter Svitter Jun 24, 2026

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

should-fix — Language inconsistency. The other docs/plan/ documents (e.g. preload-review-context.md) are written in Chinese. This document is entirely in English. If the project's doc convention is Chinese, this should be rewritten. If English is intentional for an international-facing research artifact, that's fine — but it should be a deliberate choice, not an oversight. Which is it?


## Evaluated Projects

### Tier 1: Self-Hosted AI Review (Most Relevant)

| Project | Stars | License | Multi-Agent | Self-Hosted | Webhook/CI |

@Svitter Svitter Jun 24, 2026

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

nit — Star counts (1.2k, ~11k, 33k, ~1k, 366, 57, 39) are stale-sensitive. Consider adding a note like "as of 2026-06-24" to the table header, or converting stars to a less volatile metric. Not a blocker, but readers will want to know the snapshot date.

|---|---|---|---|---|---|
| [Kodus AI](https://github.com/kodustech/kodus-ai) | 1.2k | AGPLv3 | ❌ Single agent | ✅ Docker Compose | ✅ GitHub/GitLab/Bitbucket |
| [PR-Agent (Qodo)](https://github.com/The-PR-Agent/pr-agent) | ~11k | Apache 2.0 | ❌ Single agent | ✅ Ollama | ✅ GitHub/GitLab |
| [Tabby](https://github.com/TabbyML/tabby) | 33k | Apache 2.0 | ❌ Completion-focused | ✅ Self-contained | ❌ Not review-first |

@Svitter Svitter Jun 24, 2026

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

nit — Tabby (33k stars) is primarily a code completion / IDE assistant, not a code review tool. Listing it under "Self-Hosted AI Review" is misleading. It's listed as ❌ Not review-first which is correct, but it probably belongs in a separate "adjacent tools" note rather than the Tier 1 table. A reader skimming the table might assume Tabby is a direct competitor.


### Tier 2: GitHub Actions (Simple Single-Agent)

| Project | Stars | Last Updated | Notes |
|---|---|---|---|
| [villesau/ai-codereviewer](https://github.com/villesau/ai-codereviewer) | ~1k | Dec 2023 | Stale, model deprecated |
| [cirolini/genai-code-review](https://github.com/cirolini/genai-code-review) | 366 | May 2024 | Near-stale |
| [snarktank/ai-pr-review](https://github.com/snarktank/ai-pr-review) | 57 | Active | Claude-only, early stage |
| [augmentcode/review-pr](https://github.com/augmentcode/review-pr) | 39 | Nov 2025 | Requires Augment API (closed-source) |

### Tier 3: Rule-Based (Non-LLM)

| Project | Stars | Notes |
|---|---|---|
| SonarQube Community | ~10.3k | 21 languages, rule-based, no LLM |
| Semgrep | N/A | Custom rules, security-focused |
| CodeQL | N/A | Requires GitHub Advanced Security for private repos |

### Tier 4: SaaS (Not Open-Source)

| Project | Pricing | Notes |
|---|---|---|
| CodeRabbit | $12/user/month | No open-source self-hosted option |
| Augment Cosmos | Enterprise | Closed-source |

## Key Findings

### What opencode-actions does that others don't

1. **Multi-agent parallel review**: Spawns N specialized reviewer sessions (quality, security, performance, architecture, regression-test, feature-missing, test-value, spec-coverage) concurrently via OpenCode SDK
2. **Coordinator synthesis**: A coordinator session reads all reviewer outputs and produces a deduplicated, synthesized comment
3. **Model flexibility**: Works with any model supported by OpenCode (DeepSeek, GLM, Claude, GPT, local via LiteLLM)
4. **Zero infrastructure**: Runs entirely within GitHub Actions — no web service to maintain
5. **Customizable personas**: Built-in reviewer personas with extensible configuration

### What alternatives offer

- **Kodus AI**: Best alternative for single-agent review with learning/memory, but requires maintaining a Docker Compose service and lacks multi-agent parallelism
- **PR-Agent**: Most popular open-source option, but legacy project with known configuration bugs (#2098, #2083) blocking local model deployment
- **CodeRabbit**: Best overall product, but closed-source SaaS

### Kodus AI Community Edition Limitations

| Feature | Community (Free) | Teams ($10/dev/month) |
|---|---|---|
| Kody Rules (custom rules) | Up to 10 | Unlimited |
| Active plugins | Up to 3 | Unlimited |
| Priority queue for Kody Agents | ❌ | ✅ |
| Engineering Metrics | ❌ | ✅ |

## Conclusion

opencode-actions' multi-agent parallel review architecture is unique in the open-source ecosystem. The project should be maintained rather than replaced. No existing alternative provides:
- Parallel specialized reviewer personas
- Coordinator synthesis of multi-agent findings
- Zero-infrastructure GitHub Actions deployment
- Full model flexibility via OpenCode SDK

## Recommendation

Continue maintaining opencode-actions. The multi-agent parallel review capability is a genuine differentiator with no open-source equivalent.
Loading