The Agent Skills Directory

[SAFE]: The skill's primary content is instructional, providing developers with best practices for consolidating log events into a single, context-rich 'wide event'.
[DATA_EXPOSURE]: The skill recommends including environment characteristics in logs, such as commit_hash, version, and region (found in rules/context.md and rules/structure.md). While standard for production observability, users should ensure that automated environment capture does not include sensitive secrets or private configuration data.
[INDIRECT_PROMPT_INJECTION]: The guidelines recommend ingesting untrusted data from request headers and bodies into the log context.
Ingestion points: rules/wide-events.md and SKILL.md demonstrate capturing c.req.json() and c.req.header('user-agent') for inclusion in the log event.
Boundary markers: No specific delimiters or safety warnings are suggested for the logged request data.
Capability inventory: The skill is purely instructional and does not provide executable scripts with subprocess, network, or file-system capabilities.
Sanitization: No sanitization or escaping of the request data is mentioned before it is added to the wideEvent object.

logging-best-practices