observability
Pass
Audited by Gen Agent Trust Hub on Oct 5, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill serves as a guide for implementing observability in software development. It promotes standard engineering principles such as structured logging, health checks, and failure diagnostics.
- [DATA_EXPOSURE]: The instructions explicitly warn against logging sensitive information such as secrets, tokens, or PII (Personally Identifiable Information), which is a critical security best practice.
- [PERSISTENCE]: The skill mentions 'persisted failure state', but this refers to writing diagnostic logs and error statuses to specific project directories (e.g.,
.gsd/runtime/) to assist with debugging, rather than malicious persistence mechanisms meant to survive system reboots or maintain unauthorized access. - [INDIRECT_PROMPT_INJECTION]: While the skill involves logging data that may originate from external sources (e.g., API responses), it advocates for structured JSON logging and provides a framework for how the agent should instrument its own code. It does not introduce new attack surfaces for prompt injection beyond standard application logging practices.
Audit Metadata