verify
Warn
Audited by Gen Agent Trust Hub on Jul 12, 2026
Risk Level: MEDIUMREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill resolves and executes shell commands retrieved from the body of a GitHub issue labeled 'tunan:config'. This allows for dynamic code execution based on remote configuration.
- [COMMAND_EXECUTION]: The skill uses a Bash tool to run the resolved commands, facilitating general project tasks like building and testing.
- [DATA_EXFILTRATION]: The skill fetches data from GitHub issues and pull requests, including descriptions and comments, using the GitHub CLI. While fetching from a well-known service like GitHub is generally safe, it exposes external content to the agent context.
- [PROMPT_INJECTION]: The skill processes untrusted input from GitHub issue comments when evaluating acceptance criteria, creating a surface for indirect prompt injection. Ingestion points: GitHub issue and PR bodies/comments via 'gh' CLI. Boundary markers: The skill parses fenced 'yaml' blocks and looks for specific 'tunan:gate' markers. Capability inventory: Access to the Bash tool for command execution and delegation to other skills. Sanitization: The skill relies on YAML parsing but lacks explicit sanitization of the retrieved commands before shell execution.
Audit Metadata