skills/lyndonkl/claude/audit-drift/Gen Agent Trust Hub

audit-drift

Pass

Audited by Gen Agent Trust Hub on Sep 3, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external text content (Substack posts) to evaluate them against thematic promises. This introduces a surface for indirect prompt injection where adversarial content in a post could attempt to manipulate the auditing logic.
  • Ingestion points: Content from assigned Substack posts (referenced in SKILL.md workflow).
  • Boundary markers: No explicit delimiters or instructions for the agent to ignore embedded commands are present.
  • Capability inventory: The skill is limited to internal classification and tagging; it does not possess file system write access, network capabilities, or subprocess execution.
  • Sanitization: No sanitization or filtering of the post content is described.
  • [NO_CODE]: This skill consists entirely of natural language instructions and logical workflows for the agent to follow. It does not include scripts, binaries, or automated command-line execution patterns.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 3, 2026, 02:36 PM
Security Audit — agent-trust-hub — audit-drift