implement
Pass
Audited by Gen Agent Trust Hub on Mar 24, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted data from external sources.
- Ingestion points: The agent fetches content from GitHub issues using
gh issue viewand reads implementation plans fromtmp/issues/<issue番号>/plan.md. - Boundary markers: The instructions lack explicit delimiters or safety warnings to ignore instructions embedded within the issue description or the plan file.
- Capability inventory: The skill has access to powerful tools including
Bash,Write,Edit, andTask, which allow it to execute shell commands and modify the filesystem. - Sanitization: There is no evidence of sanitization or validation of the retrieved content before it is used to guide the implementation process.
- [COMMAND_EXECUTION]: The skill utilizes the
Bashtool to perform implementation tasks and execute tests. While intended for legitimate development, this capability could be exploited if an attacker-controlled issue or plan file contains malicious commands that the agent then executes.
Audit Metadata