| name | cloudflare-ops |
| description | Guidance on how to observe Cloudflare resources. Use this before asking the Cloudflare MCP about workers, logs, or AI Gateway behavior. |
Cloudflare ops
Guidance for how to use the Cloudflare MCP for this project.
- For historical
aig LLM traffic, ask about AI Gateway logs for gateway llms.
- For worker runtime/debug logs on
aig or llms, ask about Workers Observability.
- For custom metrics only, ask about Analytics Engine datasets
aig_events and llms_usage.
- Only use worker tails when you explicitly need a live stream.
- For actual live tails, use Wrangler from the worker directory:
cd wrangler/aig && wrangler tail --format prettycd wrangler/llms && wrangler tail --format prettycd wrangler/hooks && wrangler tail --format pretty
Important repo facts
aig forwards provider traffic through AI Gateway llms.aig, llms, and hooks are the relevant worker names in this repo.aig_events is not a full log store; it only has lightweight custom datapoints.llms_usage is usage analytics, not primary request logging.- Tailing workers is usually the wrong first move when you want historical logs.
Rule of thumb
- Want request history for
aig? → AI Gateway (llms)
- Want edge/runtime logs? → Workers Observability
- Want counters/latency/tokens/custom metrics? → Analytics Engine
- Want live output right now? → tail
