We Built an On-Call Agent That Failed the Right Way — Memory Can Learn the Wrong Lesson
Similar-Split7292 · reddit · 2026-09-29
A team of four spent a weekend building Provenance, an on-call agent that investigates incidents in a simulated company, learns from past incidents via Hindsight, and verifies its memories before acting.
The author's key insight wasn't making the agent succeed, but designing an incident where it fails for the right reason: s16 was a poisoned queue message disguised as a memory leak; the agent followed a familiar pattern, suggested a rollback, and missed the real cause.
Takeaway: an agent can have memory and still learn the wrong lesson. The write-up covers building the incident, designing realistic investigation evidence, and how agent memory should actually be used.
More from coding & agent
- Millions of person-hours wasted building AI harnesses, erased by new model releases — sebpaquet · 2026-09-29
- Unverified Claim: Anthropic Engineers Share All Claude Sessions, Teammates Can Steal Each Other's Tasks — YouJiacheng · 2026-09-29
- Open-source self-driving sim repo auto-galleries 38 demos, adds Claude Code PR review skill — 4310sy · 2026-09-29
- Celesto: open-source persistent microVM computers for AI agents, boots in 500ms — aniketmaurya · 2026-09-29
- A practical guide to adopting AI in your organization: management skills over prompt tricks — chribonn · 2026-09-29
- RSI Arena: 8 AI agents get 1,000 GPU-hours each to train a better model live — my_cat_can_code · 2026-09-29