Назад към всички

openclaw-model-failover

// OpenClaw plugin: auto-detect rate limits/quota/auth failures and switch sessions to fallback LLMs. Prevents 'API rate limit reached' loops by patching pinned session models.

$ git log --oneline --stat
stars:1,933
forks:367
updated:March 4, 2026
SKILL.mdreadonly
SKILL.md Frontmatter
nameopenclaw-model-failover
slugopenclaw-model-failover
descriptionOpenClaw plugin: auto-detect rate limits/quota/auth failures and switch sessions to fallback LLMs. Prevents 'API rate limit reached' loops by patching pinned session models.

openclaw-model-failover (OpenClaw Plugin)

This is an OpenClaw Gateway plugin, not an agent skill.

It mitigates outages by automatically switching the model used for a conversation when it detects:

  • Rate limits / quota exhaustion (HTTP 429, RESOURCE_EXHAUSTED)
  • Auth/scope failures (e.g. OpenAI: missing api.responses.write)

Install

Option A: ClawHub

clawhub install openclaw-model-failover

Option B: From repo (dev)

openclaw plugins install -l ~/.openclaw/workspace/openclaw-model-failover
openclaw gateway restart

Configure

Set model fallback order (example):

  • Anthropic -> OpenAI -> Google

See README for the full config block.