my-knowledge

Pass

Audited by Gen Agent Trust Hub on Aug 25, 2026

Risk Level: SAFE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests and analyzes the full conversation history, which qualifies as untrusted data from a security perspective.
  • Ingestion points: Phase 1 of the skill scans the complete session history (SKILL.md).
  • Boundary markers: No explicit delimiters are used to separate user-provided history from the assessment instructions.
  • Capability inventory: The skill is strictly limited to text output. It contains explicit prohibitions against writing to disk, calling external services, or executing CLI commands (SKILL.md, Constraints section).
  • Sanitization: No explicit sanitization of history content is performed.
  • [DATA_EXPOSURE_AND_EXFILTRATION]: While the skill accesses sensitive session data (the conversation history), it includes instructions that prevent exfiltration.
  • Evidence: The skill specifies: 'write no files', 'call no services', and 'deliver your entire assessment as an inline response and then return to silence.'
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 25, 2026, 09:14 PM
Security Audit — agent-trust-hub — my-knowledge