Multi-model automatic fallback system
ReviewAudited by ClawScan on May 10, 2026.
Overview
This is a purpose-aligned documentation-only model fallback skill, but it is incomplete and may route your prompts or documents to backup model providers you configure.
Before installing, verify that you actually have a reviewed implementation for the referenced script/config, choose only trusted model providers, and understand that automatic fallback can send the same user request or document to a backup provider.
Findings (4)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Manual commands may fail or could accidentally run an unrelated local script if one exists at that path.
The supplied manifest contains only SKILL.md and no scripts or config files, so this referenced helper is not part of the reviewed package.
/scripts/model-fallback.sh --force-next
Do not run unverified helper scripts; confirm the intended implementation and config files are present and reviewed before enabling fallback.
Your agent may retry requests on a different provider, which can affect behavior, cost, latency, and output quality.
Automatic provider switching is central to the skill, but it changes runtime model selection without a per-request prompt once configured.
The skill automatically handles model failures. No explicit calls needed.
Enable this only if you want automatic fallback, and review the fallback chain, timeouts, retry counts, and reset behavior.
Users may need to supply model-provider credentials even though the metadata does not make that dependency explicit.
The documentation expects provider API keys, while the registry metadata declares no required credentials or environment variables.
Check API keys are valid
Use only the minimum provider credentials needed, store them through the platform’s normal secret mechanism, and avoid pasting keys into chat or config files unless intended.
Prompts, files, or document contents submitted to the agent may be processed by backup providers, not only the primary provider.
Requests may be routed from the primary model to alternate external providers such as MiniMax, Kimi, or Zhipu.
automatically switches to backup models in a predefined priority order
Configure only trusted providers, review their data-retention policies, and avoid enabling fallback for sensitive data unless those providers are approved.