buildConfirmed12 publishers OpenAI disclosed the incident on September 16 under a framework it polices itself. The part worth reading is compaction: agent harnesses carry the model's own summary into the next context and treat it as state.
Perspective Coverage
12 publishers
- Builder
- Builder 41%
- Operator
- Operator 37%
- Investor
- Investor 22%
Reality
- Evidence70
- Adoption
- Not scored
- Hype gap+25
- Incentives70
- Confidence66
buildOne report1 publisher Codex gives the model one bounded turn to pick where its own context gets cut
A design reading of one Codex commit shows the runtime holding a 90% compaction budget and a 95% hard guard, with the model allowed a single bounded step in between to call new_context at a cleaner point.
Reality
- Evidence48
- Adoption
- Not scored
- Hype gap+14
- Incentives22
- Confidence55
buildOne report1 publisher myc's PreCompact hook writes the session to disk before the summary drops the reason
myc is an MIT-licensed memory layer for coding agents that installs into Claude Code's PreCompact event, writes the raw session to disk before it prints anything back, and hides the decisions it extracts until a human confirms them.
Reality
- Evidence38
- Adoption10
- Hype gap+12
- Incentives65
- Confidence45
buildOne report1 publisher Seven of sixteen compacted agent sessions answered from the summary instead of re-reading the file
Compaction leaves behind a summary that can outvote the file it summarised. A small hands-on trial on configuration files found the model serving stale values out of that summary in nearly half of the compacted runs.
Reality
- Evidence45
- Adoption
- Not scored
- Hype gap−5
- Incentives20
- Confidence50
buildOne report1 publisher Uber's AI cost per session fell 18 points further than its cost per request
The gap implies each session now uses roughly a quarter fewer requests, so the compaction threshold and the prompt cache in Uber's own harness are carrying part of a saving usually credited to cheaper open models.
Reality
- Evidence45
- Adoption62
- Hype gap+26
- Incentives58
- Confidence48
buildOne report1 publisher clio-coder advertised two compaction settings its orchestrator never read
The project's own audit traced the read path with file:line receipts and closed the gap in about 90 minutes, while a same-week Codex issue shows the harder version, where the key is read and quietly replaces defaults.
Reality
- Evidence64
- Adoption24
- Hype gap+22
- Incentives55
- Confidence58
buildOne report1 publisher Compaction that cut tool output 38.4% pushed the bill up 6.8%
Provider prompt caches bill a reused prefix at roughly a tenth of input, so an agent that rewrites its own history to save tokens forfeits the discount and pays to re-prefill everything ahead of the edit.
Reality
- Evidence30
- Adoption
- Not scored
- Hype gap+34
- Incentives35
- Confidence34