commit-security-scan

Pass

Audited by Gen Agent Trust Hub on Sep 16, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill retrieves and analyzes untrusted code diffs and pull request data via git diff and gh pr diff as described in Step 2 and Step 4. This ingestion creates a vulnerability surface where an attacker could embed malicious instructions within code comments or string literals to manipulate the LLM's reasoning or suppress security findings. \n
  • Ingestion points: Step 2 (commands fetching untrusted code changes from commits and PRs).\n
  • Boundary markers: The instructions do not define clear delimiters or include explicit warnings for the agent to ignore instructions embedded within the code being analyzed.\n
  • Capability inventory: The skill can execute shell commands (git, gh), read repository files, and write to security-findings.json.\n
  • Sanitization: No sanitization or filtering is applied to the code content before it is passed to the LLM for analysis.\n- [COMMAND_EXECUTION]: The skill uses standard command-line utilities such as git and gh to perform its primary function of retrieving code changes. These commands are used for their intended purpose in a developer-oriented workflow. Documentation in analysis-examples.md includes strings like 'rm -rf /' in exploit scenarios, but these are provided as pedagogical examples of vulnerabilities and are not executed by the skill.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 16, 2026, 04:30 PM
Security Audit — agent-trust-hub — commit-security-scan