The Agent Skills Directory

[SAFE]: The skill contains no malicious code, obfuscation, or unauthorized data access patterns.
[DATA_EXFILTRATION]: The logging implementation specifically includes a redaction configuration for the pino logger, preventing the accidental logging of sensitive data such as api_key, email, phone, and Authorization headers.
[COMMAND_EXECUTION]: While the skill frontmatter requests permissions for kubectl and curl, no bash commands or subprocess executions are present in the provided source files.
[PROMPT_INJECTION]: The skill processes data from external API responses, creating a potential surface for indirect prompt injection, although the risk is minimal given its primary use in metrics and logging.
Ingestion points: src/lib/apollo/instrumented-client.ts (response headers and error objects), src/routes/health/apollo.ts (health check response).
Boundary markers: Absent.
Capability inventory: kubectl, curl, Read, Write, Edit (from frontmatter).
Sanitization: None; raw error messages and header values are passed directly to logging, tracing, and health check endpoints.

apollo-observability