AI agent fixed the bug, but what else did it change?
Trail of Bits announced two open‑source plugins, patching-bugs and regression-hunt, built with OpenAI under the joint "Patch the Planet" initiative. The tools let engineers delegate bug reproduction, fix generation, and regression checking to AI agents, while giving maintainers control through threat‑model and patching‑guideline files.
Key points
- Regression‑hunt cuts false‑positive regression findings by 25% versus a baseline Codex agent (95% CI 14‑38%).
- Evaluation found 21 confirmed regressions among 58 candidates (≈36% precision) from over 130 merged PRs.
- Plugins were co‑created by Trail of Bits and OpenAI, using strict schemas, live PoCs, and threat‑model integration.
In internal evaluations, regression‑hunt reduced false‑positive regression findings by 25% compared with a baseline Codex agent, with a 95% confidence interval of 14% to 38%. Across more than 130 merged pull requests, the system identified 21 confirmed regressions from 58 candidate findings (about 36% precision). The plugins enforce strict input/output schemas, live proof‑of‑concept runs, and modular steps that can be used independently. Future work will add configurable rules, broader evaluation metrics, and deeper analysis of agent trace data.
The release aims to standardize patch development and regression validation, helping projects move from bug reports to reliable patches faster while keeping security and quality safeguards in place.
AI agent fixed the bug, but what else did it change?
blog.trailofbits.com · 21 October 2026
Loading the full article…
This text was published by blog.trailofbits.com and written by "Anish Naik, Marcelo Morales". It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
Coverage and discussion
1source- Hacker News discussion · 42 pointsnews.ycombinator.com
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Agents & Tools
All →- Aws introduces Adaptive agentic intelligence framework with Strands Agents · 1 src
- Opposable lets AI tap and type on Android phones and iPhones · 1 src
- Walmart launches Agent Gateway for AI shopping agents · 3 src
- Hugging Face releases TorchNeuron beta for easier Hugging Face models on AWS Trainium · 2 src
- LTSharpe and Claude AI boost Oblivion remastered performance by up to 68% · 2 src
Comments
via GitHub Discussions