- name
- rate-limit-guard
- description
- Prevent 429 retry loops and wasted tokens. Use before expensive API calls or after hitting rate limits. Enforces preflight checks, backoff, and context shrinking.
Rate Limit Guard
Before Expensive Calls
Check: provider/model, recent failures, concurrency, context size.
On 429
- Set concurrency to 1.
- Stop parallel retries.
- Shrink context.
- Send one minimal probe.
- Back off (exponential).
Rules
- Never retry a giant request unchanged.
- Never fallback-spam premium providers.
- Batch corpus work in 10–20 item chunks.
Output
Next action: proceed / recovery mode / pause.