The Agent Skills Directory

[DATA_EXFILTRATION]: The skill implements logic to collect and transmit structured JSON logs, trace spans, and error stack traces to an external collector service using Cloudflare Service Bindings (env.LOG_COLLECTOR). This behavior is consistent with the skill's primary purpose of providing observability for edge applications.
[INDIRECT_PROMPT_INJECTION]: The skill contains a vulnerability surface where untrusted data from incoming HTTP request headers is ingested and processed.
- Ingestion points: The RequestTracer class ingests the x-trace-id header from incoming requests in SKILL.md. Error logging also captures the request URL and method.
- Boundary markers: No explicit boundary markers or instructions to ignore embedded content in headers are present in the template code.
- Capability inventory: The skill has the capability to perform network requests via Service Bindings (env.LOG_COLLECTOR.fetch) to ship collected data.
- Sanitization: The code provides a fallback to crypto.randomUUID() if the header is missing, but does not explicitly sanitize the content of the x-trace-id header or other request metadata before logging.

observability