audit-project-context

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes data from the repository via git diff and file reads from /project-context. An attacker could potentially influence the agent's behavior by embedding instructions in the code or documentation being audited. * Ingestion points: git diff output and repository files (SKILL.md, CONTRIBUTING.md, and code files). * Boundary markers: The skill defines specific classification rules (Keep, Cut to a link, Cut) to guide the agent's decision-making. * Capability inventory: The skill performs file reads, file writes (making cuts), and executes git commands. * Sanitization: No explicit sanitization or escaping of the repository content is specified.
  • [COMMAND_EXECUTION]: The skill utilizes git diff to compare versions. While $ARGUMENTS are used to define the base reference, this is standard functionality for repository maintenance tools.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 07:27 AM
Security Audit — agent-trust-hub — audit-project-context