🚀 Meet AIDO
The AI companion for the pull requests your AI writes.
AI agents — Copilot, Claude Code, Cursor — are opening more and more pull requests, and a human still has to understand code they didn't write. Aido keeps that human in the loop: when an AI-authored PR lands, it can automatically explain, summarize, review, and document the change. And you can run those same commands on any PR or issue on demand — just comment aido <command>.
Vibe-code all you want — with reviewers who actually care about the code. One companion for the whole review lifecycle (review, summarize, explain, document, test, triage), with Gemini, ChatGPT, Claude, or any OpenAI-compatible model (Mistral, DeepSeek, …), right inside GitHub Actions. Bring your own key; no third-party data processor.
⏱️ 60-second start
- Add a
GEMINI_API_KEYrepo secret (free key — Settings → Secrets and variables → Actions). - Copy
examples/remote/aido.yml→.github/workflows/aido.yml(one file). - Comment
aido reviewon any PR.
That's it — Aido replies right in the PR. → Full install options (copy-based, GitHub Action step, other providers).
👀 See it live (no install)
Real Aido comments on demo PRs — click any command to see the full, rendered output:
| Command | What it does | Live demo |
|---|---|---|
aido review |
Multi-persona review + digest, inline applyable suggestions | PR #64 |
aido suggest |
Concrete improvements & small refactors | PR #65 |
aido test |
Test plan, coverage gaps, follow-ups | PR #66 |
aido explain |
Developer-focused step-by-step walkthrough | PR #67 |
aido summarize |
High-level summary for stakeholders | PR #68 |
aido docs |
Draft/augment documentation | PR #69 |
🔬 One PR, three models: PR #106 reviews the same diff with Claude, ChatGPT, and Gemini side by side — same personas, different provider. A concrete look at what "bring your own model" buys you: each catches the core bugs, but they differ on depth and which second-order issues they flag.
Sample: aido review catching real bugs (from PR #64)
This PR introduces several common Python anti-patterns and security vulnerabilities, including a mutable default argument, SQL injection, and a bare
exceptclause.Recommendation: Request changes
Faceted Notes:
- Security: Critical SQL injection vulnerability; bare
exceptmasks errors.- Correctness: Mutable default argument leads to unexpected state; missing input validation.
- Maintainability: "Magic numbers" reduce readability.
- QA/Testing: Bare
exceptswallows all errors, making failure modes hard to test.
Code fixes arrive as inline "Apply suggestion" buttons on the diff — not dumped in the comment body.
✨ What you get
- 🤖 Auto-companion for AI-authored PRs — when Copilot / Claude Code / Cursor open a PR, Aido runs automatically (explain + summarize by default; review/docs/test opt-in). → details
- ⚡ On-demand on any PR or issue —
review,summarize,explain,docs,suggest,test,triage. → command reference - 🧩 Persona-guided reviewer with applyable inline suggestions (robust validation). → personas
- 🔌 Multi-provider, bring-your-own-key — Gemini, ChatGPT, Claude, or any OpenAI-compatible endpoint (Mistral, DeepSeek, Kimi, Grok, …). No third-party data processor. → providers
- 🗓️ Weekly "what shipped" digest — a scheduled summary of merged PRs, posted as an Issue or Discussion. → details
- 📦 One-file install from a pinned release tag; upgrading is a one-line bump.
❤️ Why Aido?
As more of your PRs are written by AI, Aido makes sure a human still understands them — automatically, and on demand. Open-source, configurable, and team-friendly: catch issues early and keep everyone in the loop, without leaving GitHub and without sending your code to a third-party processor.
Several providers have free tiers, so you can run it for free on GitHub Actions minutes with your own key. → Free to run
📚 Documentation
- Installation — requirements, remote/copy install, GitHub Action step, free providers
- Commands — full command list, summarize-vs-explain, tips
- Configuration — config keys, per-command configs, diff size, bring-any-model
- Persona reviews — how the reviewer works, controls, persona packs
- Auto-run & digest — auto-run on AI PRs, the weekly digest
- Architecture — how it works, data handling, reliability
- Security — reporting + adopter security posture
Happy shipping! ✨
Note
AI can be incorrect, rate-limited, or change without notice, and prompts/code may be sent to external providers. Always verify reviews and suggestions, keep a human in the loop, and never wire Aido's output to auto-merge. See Architecture → data handling and SECURITY.md.
