Agents turn conservative as context grows: edits shrink from 150 to 30 lines
dbreunig · x · 2026-10-16
Analyzing agent trajectories, researchers found agents grow more conservative as their sessions lengthen. On the AHC014 competitive programming task, per-proposal code edits shrank from 150 lines to 30, and thinking tokens fell from 28k to 7k per turn as context grew. After each "forget" (context reset), both metrics jumped back up — suggesting context accumulation itself drives the plateau.
More from coding & agent
- Google-led paper uses sequential auctions to organize decentralized LLM agent workflows — _onionesque · 2026-10-16
- Since OpenAI's math release, GitHub contributors keep driving down the integer-multiplication kappa — xennygrimmato_ · 2026-10-16
- Pattern MCP makes coding agents check existing components before writing new code — DonR954 · 2026-10-16
- Zero-keyframe procedural spider walks walls and ceilings using pure raycasts — yshan2u · 2026-10-16
- OpenCode v1.19.2 keeps 1.x installs off the 2.x update path — opencode-agent[bot] · 2026-10-16
- NVIDIA App update turns on AI agent control by default — here's how to switch it off — ssh4net · 2026-10-16