NYC

fetch-unresolved-comments

Fail

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: HIGHPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • Prompt Injection (HIGH): The skill is highly vulnerable to Indirect Prompt Injection (Category 8).
  • Ingestion points: The skill fetches comments[].body from the GitHub GraphQL API, which contains arbitrary text written by any user with access to the PR.
  • Boundary markers: There are no instructions in SKILL.md to wrap the comment data in delimiters or warn the agent to ignore embedded instructions.
  • Capability inventory: The agent is expected to "Parse the JSON output" and likely use the content to summarize feedback or guide code changes, providing a clear path for an attacker to influence agent behavior.
  • Sanitization: No sanitization or filtering of the comment body is mentioned before it is processed by the agent.
  • Command Execution (MEDIUM): The skill instructions require the execution of shell commands, including gh pr view and uv run python. This introduces a dependency on the local environment and the integrity of the files located at .claude/skills/fetch-unresolved-comments/.
  • Data Exposure (LOW): The skill accesses the user's GitHub authentication token using gh auth token. While this is a standard practice for GitHub integration, it involves handling sensitive credentials that could be exfiltrated if combined with other malicious patterns.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 16, 2026, 10:54 AM