Context compression cut token use, but broke AI agent reliability
PepperWestern2263 · reddit · 2026-07-23
A Reddit user says context compression for AI agents reduced token usage, but also broke reliability in subtle ways.
They split context into three categories:
- Disposable context: repeated search results, duplicated docs, long file listings, verbose logs. Good candidates for filtering or summarization.
- Load-bearing context: exact error messages, file paths, line numbers, patch anchors, test names, acceptance criteria. Even small rewrites can break the next step.
- Machine-consumed context: JSON, shell output, CSV, patches, or anything parsed by another tool. This is where compression caused the most surprising failures.
The main lesson: the goal should not be the smallest possible context, but the smallest context that still preserves the evidence and interfaces needed for the next action.
The author asks how others handle this in production, and whether teams use explicit no-summarize rules or rely on retrieval.
More from coding & agent
- Spring AI 2.1.0-RC1 adds tool-call confirmation, pausing agent loops before real-data access — therealdanvega · 2026-10-26
- Morgan Stanley and US banks block third-party AI agents over credentials — thione · 2026-10-26
- METR red-teams OpenAI's agent monitoring, finds bypasses and gaps — thione · 2026-10-26
- OpenAI: coding agents weakened safety monitoring, one lied about it — thione · 2026-10-26
- ConvFill: A 0.6B Talker Buys Time While a Large Reasoner Whispers the Real Answer — andimarafioti · 2026-10-26
- 250 Builders Hack With Codex at Berlin Community Hackathon Ahead of OpenAI DevDay — paw_lean · 2026-10-26