An LLM council that reviews your coding agent's every move
Sage is an LLM-based reviewer that critiques coding agents’ plans, responses and code in real time.
- for
- Developers using AI coding assistants who want automated, contextual reviews.
- pricing
- open source
- license
- open source
works withClaude CodeOpenAI Codex SDK
Key features
6 features of Sage
- Critique cards — Generates structured verdicts, alternatives and ready-to-paste messages for each agent response.
- Plan-level review — Analyzes the agent’s proposed plan before any code is written to catch design flaws early.
- Model council — Runs a Codex-based reviewer and will later combine multiple models (Claude, Gemini, Grok) for cross-checking.
- Tool-enabled critiques — Can search the web, inspect the codebase and make tool calls during reviews.
- Terminal UI — Provides an interactive Ink-based interface that runs alongside Claude Code.
- Hooks integration — Automatically configures Claude Code hooks to monitor session events and trigger reviews.
Use cases
- Validate an AI agent’s design plan before committing code
- Cross-check responses from Claude Code with a Codex reviewer
- Get automated suggestions and alternatives during a coding session
- Detect reasoning or planning errors that diff-only reviewers miss
Sage vs alternatives
 Sage |  Fine |  ReviewPhin |  SuperAGI Cloud |  Pocketenv | |
|---|---|---|---|---|---|
| Best for | LLM council that reviews AI coding agents | Building AI agents | Self-hosted code review | Cloud-hosted autonomous agents | Agent sandbox runtime |
| Pricing | Open source | Free | Free | Free | Free |
| DevHunt upvotes | 3 | 75 | 12 | 43 | 73 |
| Launched | Dec 2025 | Jan 2023 | Jul 2026 | Jan 2023 | Apr 2026 |
- Sage vs Fine: Fine focuses on building AI agents, while Sage reviews existing coding agents’ output.
- Sage vs ReviewPhin: ReviewPhin provides self-hosted code review for GitLab/GitHub, whereas Sage reviews live LLM conversations.
- Sage vs SuperAGI Cloud: SuperAGI Cloud runs autonomous agents in the cloud; Sage is a local reviewer that augments a single coding agent.
- Sage vs Pocketenv: Pocketenv offers a sandbox runtime for agents, while Sage adds a review layer on top of the coding agent itself.
Sage FAQ
Which operating systems does Sage support?+
Sage has only been tested on macOS; support for other OSes may be limited.
How does Sage connect to Claude Code?+
On first run Sage configures Claude Code hooks; it monitors session files and triggers reviews automatically.
What models does Sage currently use?+
It currently uses the OpenAI Codex SDK and plans to add Claude, Gemini and Grok for a multi-model council.
Can Sage review sessions created before installation?+
Only sessions created after installation are visible; older sessions can be resumed with `claude --resume <id>` to make them reviewable.
Why might the terminal flicker?+
iTerm2 flickering is due to the Ink library; using the default macOS Terminal or VS Code’s integrated terminal avoids it.
Summarized by DevHunt from github.com · Sep 29, 2026. Details may change; check the official site.
