The Agent Skills Directory

[SAFE]: The skill provides implementation patterns for production observability using industry-standard libraries and frameworks. It includes templates for structured logging, distributed tracing with OpenTelemetry, and metrics collection with Prometheus.
[INDIRECT_PROMPT_INJECTION]: The skill defines middleware that ingests external data from HTTP headers for request correlation purposes.
Ingestion points: CorrelationIDMiddleware in SKILL.md reads x-request-id and x-correlation-id from request headers.
Boundary markers: None explicitly defined for these IDs, as they are treated as opaque strings for logging and tracing context within the application flow.
Capability inventory: The skill provides instructions for logging (structlog, pino), tracing (opentelemetry), and metrics (prometheus_client). It does not include subprocess execution, dynamic code execution, or file writes based on this external data.
Sanitization: The pino setup in SKILL.md includes a redact configuration to prevent logging of sensitive fields such as authorization headers, cookies, and passwords. The middleware decodes the header but does not perform complex parsing that would be vulnerable to injection.

production-monitoring