paper-bootstrap

Pass

Audited by Gen Agent Trust Hub on Sep 18, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to evaluate existing workspace files to determine a manuscript source of truth, creating a potential surface for indirect prompt injection attacks.
  • Ingestion points: The agent is directed to inspect existing drafts, notes, and manuscript source files within a project directory (SKILL.md).
  • Boundary markers: No explicit delimiters or instructions are provided to ensure the agent ignores or isolates instructions embedded within the analyzed documents.
  • Capability inventory: The skill provides a script (scripts/init_paper_layout.py) that performs directory creation and file writing. The agent is also directed to update its internal project memory based on the findings from its analysis.
  • Sanitization: The skill does not implement validation or sanitization for the content read from external files before it is used to influence the agent's state or memory.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 18, 2026, 04:20 PM
Security Audit — agent-trust-hub — paper-bootstrap