# Program Repair

Coding-agent evaluation becomes a duel in which two agents author and solve each other's tasks: over 1,500 turns, GPT-5.5 with Codex CLI repaired 98.3% of the defects it reviewed, while Claude Opus 4.8 with mini-swe-agent was the strongest task author with a 27.3% Red win rate.

Coding-agent evaluation becomes a duel in which two agents author and solve each other's tasks: over 1,500 turns, GPT-5.5 with Codex CLI repaired 98.3% of the defects it reviewed, while Claude Opus 4.8 with mini-swe-agent was the strongest task author with a 27.3% Red win rate.

The work introduces RucTangle, an agentic commit-untangling method that groups a patch's changes by development purpose, orders them, applies each commit and checks whether the code still runs, revising order first, then regrouping via the agent, and finally merging adjacent commits while preserving the patch's final code; on 131 agent-generated patches it is the only method that reconstructs every patch with all commits runnable, at a median of eight commits, whereas baselines leave 20.6%–37.4% of histories with at least one unrunnable version. The authors also present TangleEval, which has two coding agents repair 453 regression-introducing patches; adding RucTangle histories plus git bisect results raises pass@1 from 26.9% to 31.3% and from 30.8% to 36.0%.

The work introduces RucTangle, an agentic commit-untangling method that groups a patch's changes by development purpose, orders them, applies each commit and checks whether the code still runs, revising order first, then regrouping via the agent, and finally merging adjacent commits while preserving the patch's final code; on 131 agent-generated patches it is the only method that reconstructs every patch with all commits runnable, at a median of eight commits, whereas baselines leave 20.6%–37.4% of histories with at least one unrunnable version. The authors also present TangleEval, which has two coding agents repair 453 regression-introducing patches; adding RucTangle histories plus git bisect results raises pass@1 from 26.9% to 31.3% and from 30.8% to 36.0%.

Code agents solving repository-level tasks usually read only the current source tree, leaving the design decisions and compatibility constraints recorded in merged pull requests unused. Chronos distills each merged pull request into a structured experience card and links the cards in a typed graph of code, developer-intent and organizational relations, then retrieves by semantic entry plus weighted multi-hop expansion. On SWE-Bench Verified the full workflow improves SWE-Agent on all six evaluated backbones, raising mean resolution from 69.2% to 72.9% and reaching 79.8% in the strongest configuration.

Code agents solving repository-level tasks usually read only the current source tree, leaving the design decisions and compatibility constraints recorded in merged pull requests unused. Chronos distills each merged pull request into a structured experience card and links the cards in a typed graph of code, developer-intent and organizational relations, then retrieves by semantic entry plus weighted multi-hop expansion. On SWE-Bench Verified the full workflow improves SWE-Agent on all six evaluated backbones, raising mean resolution from 69.2% to 72.9% and reaching 79.8% in the strongest configuration.