observe-it

Pass

Audited by Gen Agent Trust Hub on Oct 23, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to analyze the existing codebase, documentation, and configuration files to discover the current observability stack. This process exposes the agent to untrusted data within the repository which could contain malicious patterns or instructions.
  • Ingestion points: Workflow step 2 in SKILL.md involves reading project documentation, environment files (.env*), and SDK imports from the local repository.
  • Boundary markers: The skill utilizes an 'Operating contract' that requires defining explicit acceptance criteria and a 'finish line' before starting work to constrain agent behavior.
  • Capability inventory: The agent has the capability to modify source code to add instrumentation and execute shell commands to verify signal emission.
  • Sanitization: Step 4 of the workflow explicitly forbids the inclusion of secrets, tokens, or raw PII in default logs and traces.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 23, 2026, 01:31 AM