implement-issue
Warn
Audited by Gen Agent Trust Hub on Jul 7, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill identifies and runs arbitrary commands (test suites, build scripts, application entrypoints) based on definitions found within the repository's configuration files (e.g.,
package.jsonscripts,Makefile,pyproject.toml). Executing these scripts allows for arbitrary code execution within the agent's environment if the repository is untrusted. - [PROMPT_INJECTION]: The skill is vulnerable to Indirect Prompt Injection through processed GitHub issue content.
- Ingestion points: Reads issue bodies and comments via
gh issue view <N> --commentsinSKILL.md. - Boundary markers: Absent. The instructions do not provide delimiters or warnings to the agent to ignore instructions embedded in issue data.
- Capability inventory: Capability to execute shell commands, modify files, and perform GitHub API operations (push code, create PRs).
- Sanitization: Absent. No filtering of issue content is mentioned.
- [COMMAND_EXECUTION]: The instructions explicitly direct the agent to run network commands (
gh,git) outside of any sandboxed shell to avoid connection errors, effectively requesting a bypass of platform-enforced security boundaries.
Audit Metadata