skills/fl-sean03/pcp/vault-operations/Gen Agent Trust Hub

vault-operations

Pass

Audited by Gen Agent Trust Hub on Jun 18, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: No malicious patterns, obfuscation, or unauthorized data exfiltration attempts were detected in the skill instructions. The functionality is focused on local data management and productivity.
  • [COMMAND_EXECUTION]: The skill uses local CLI commands to interact with its core logic, specifically invoking python vault_v2.py for tasks such as capturing data, searching, and managing project status. This is standard behavior for its documented purpose.
  • [PROMPT_INJECTION]: The skill exposes a surface for indirect prompt injection by ingesting and processing untrusted user data via the smart_capture() and smart_search() functions.
  • Ingestion points: User-provided text strings in the SKILL.md usage examples.
  • Boundary markers: None present in the documentation.
  • Capability inventory: Database writes (people, tasks, projects) and local command execution via vault_v2.py.
  • Sanitization: Not mentioned in the provided documentation. This finding is considered low risk as it is intrinsic to the skill's primary function.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 18, 2026, 01:06 PM
Security Audit — agent-trust-hub — vault-operations