git-workflow

Pass

Audited by Gen Agent Trust Hub on Mar 3, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The 'PR Fix Comments Workflow' processes external, potentially untrusted data from GitHub PR comments and issue comments via commands like 'bun run gh-tool pr threads' and 'bun run gh-tool pr issue-comments-latest'. The skill instructs the agent to 'Make the code change' based on these suggestions, creating an indirect prompt injection surface. An attacker could post a malicious comment designed to trick the agent into injecting a backdoor or exfiltrating data during the automated 'Apply Fixes' step.
  • Ingestion points: PR threads and issue comments fetched in Step 2 of the PR Fix Comments Workflow in SKILL.md.
  • Boundary markers: No boundary markers or 'ignore embedded instructions' warnings are present to distinguish comments from safe instructions.
  • Capability inventory: The skill possesses extensive capabilities including file system modification, 'git commit', 'git push', and arbitrary command execution via 'bun run check'.
  • Sanitization: No sanitization or validation of the comment body is performed before the agent attempts to interpret and apply the code suggestions.
  • [COMMAND_EXECUTION]: The skill generates shell commands by directly interpolating user-controlled variables into strings, such as 'git commit -m ""' and 'bun run gh-tool pr create --title "" --body ""'. This pattern is vulnerable to command injection if the strings are not properly escaped by the agent's tool execution environment.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 3, 2026, 08:17 AM