know-me

Pass

Audited by Gen Agent Trust Hub on Sep 6, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill establishes a workflow for persistent context by managing information within a local .memory/ directory. This pattern creates a surface for indirect prompt injection if the agent processes untrusted data and stores it in these files for later recall.
  • Ingestion points: Data is read from .memory/MEMORY.md, .memory/user-preferences.md, .memory/project-context.md, and .memory/corrections.md as specified in references/memory-operations.md.
  • Boundary markers: Absent. The instructions do not specify the use of delimiters or 'ignore' instructions to prevent the agent from following commands embedded in the stored memory.
  • Capability inventory: The skill provides an organizational strategy for information storage but does not include any scripts with executable capabilities; it instructs the agent to perform file system operations as part of its memory cycle.
  • Sanitization: Absent. While the skill includes a safety rule to avoid persisting passwords or secrets, it does not provide mechanisms to sanitize or filter potential injection attacks within the stored context.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 6, 2026, 03:10 AM
Security Audit — agent-trust-hub — know-me