gomtm-bugs-finder
Pass
Audited by Gen Agent Trust Hub on Jun 28, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it instructs the agent to process untrusted external data.
- Ingestion points: The agent is directed to analyze "latest task history" and "report files" to determine bug patterns (found in
SKILL.md). - Boundary markers: There are no specified delimiters or instructions to treat the analyzed history as data rather than executable instructions.
- Capability inventory: The agent is empowered to perform "automatic repair" and write documentation directly to the "main branch."
- Sanitization: No sanitization or validation of the history/report content is mentioned before processing.
- [COMMAND_EXECUTION]: The instructions mandate the use of Docker containers for reproduction, verification, and automated fixing of code. This gives the agent a wide latitude to execute shell commands and modify the filesystem within the execution environment.
- [DATA_EXFILTRATION]: While no explicit external network calls are defined, the skill grants access to sensitive project metadata, including task history, developer intent, and code logic, which represents a data exposure risk if the agent is further compromised.
Audit Metadata