A Budget That Resets on Every Retry Isn't a Budget: Scope Agent Limits to the Task
Key-Register-1969 · reddit · 2026-10-31
A developer outlines three recurring failure modes sharing the same hole: a retrieval cap a reworded query resets, an API budget a fresh root run resets, and a retry loop an error workflow re-triggers. Limits scoped to a single run catch bursts but miss slow repeats that look like new tasks. He proposes scoping limits to the task: a task identifier that survives across runs and retries, a repeat counter keyed to task+action+inputs with a decay window long enough for minutes-apart repeats, and error paths that inherit remaining budget instead of minting a new one. The hard part is deciding when a retry becomes genuinely new work — expiry and reset conditions should tie to real change, not new run IDs.
More from coding & agent
- Curtis Yarvin proposes the "self-driving computer" where every app builds itself — ZeroStateReflex · 2026-10-31
- Non-dev vibe-codes a bike showroom's full management app, now asked for a voice agent — SaiTarunRonanki · 2026-10-31
- Designing agent evals: breaking timestamp ties with revisions without order dependence — jylusdev · 2026-10-31
- Self-Credit: self-generated credit assignment boosts long-horizon LLM agents by up to 8.5 points — gregd_nlp · 2026-10-31
- herdr hits YC and 32K followers in 5 months with durable terminal runtime for coding agents — Scobleizer · 2026-10-31
- Your AI coding agent is not your coworker yet: who owns the architecture? — Obaid_BLM · 2026-10-31