ai-jury
Multi-agent CLI that reviews diffs, PRs or issues via debating agents into one verdict.
Tested on Linux
Use it
Linux x86_64
You need first
- Runtimes:
python>=3.11 - Keys or accounts: ANTHROPIC_API_KEY, OPENAI_API_KEY
Install ai-jury via uv tool
uv tool install ai-jury
Check that it works
jury --versionThe output should contain “jury 1.28.0”.
First thing to try
jury --mockRun jury --mock to see a sample review with mock agents.
What it does
- Review pull requests, diffs or issuesTested
How to use it
- Run
jury --mockto see a sample review. - Read the findings, each listed with severity, file, line and the agents that raised it.
- Run
- Have agents debate and check each other's findingsTested
How to use it
- Run
jury --mock --rounds 2to add a cross-examination round. - Look for the "Round 2 — cross-examination" section in the output.
- Run
- Reach a verdict through a chair or a voteTested
How to use it
- Run
jury --mock --rounds 1 --decision voteto decide by panel vote. - Check the output for the ballot count behind the verdict.
- Run
- Stream output live or as a transcriptTested
How to use it
- Run
jury --mock --transcriptto get the full transcript. - Find the verdict under "Decision — verdict & reasoning".
- Run
How far pickfit got
Linux x86_64
- Installs
- Starts
- 4 of 4 features work
- Works offline
Good for
- review a GitHub PR
- review a diff file
- review an issue for completeness and clarity
- post review verdict back as PR comment
- reach verdict by panel vote instead of chair
How pickfit tested it
Each run starts from a clean machine and follows the steps above. How pickfit verifies
Linux x86_64
Its main function works
- Replayed from scratch
- Works offline
- Checked on 30 Oct 2026
- Version 1.28.0