managing-up

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill manages persistent context by instructing the agent to save and read manager profiles and organizational data in the .agents/ directory, creating a vulnerability surface where instructions embedded in a saved profile could influence the agent in future interactions.
  • Ingestion points: Context is read from .agents/em-context.md and .agents/reports/[name].md during initialization and updates.
  • Boundary markers: Absent; the instructions do not define delimiters or provide warnings to ignore potential instructions within the stored data.
  • Capability inventory: The skill utilizes file read and write capabilities for context persistence.
  • Sanitization: Absent; there is no requirement for validating or filtering user-provided context before it is saved to the local directory.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 05:46 AM
Security Audit — agent-trust-hub — managing-up