build-it
Pass
Audited by Gen Agent Trust Hub on Apr 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the dynamic context injection syntax (
!) to execute shell commands (echo,cat) at load time to display path and version information. It also instructs the agent to run CLI tools includinggh issue viewandtsc --noEmit.\n- [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection as it processes specifications from untrusted sources (GitHub issues, conversation history, and pasted text). The instructions command the agent to treat the specification as a 'contract' and implement everything within it, which could lead to the execution of malicious instructions embedded in the specification during the implementation phase.\n - Ingestion points: GitHub issue bodies, conversation context, and user-pasted text in Phase 3.\n
- Boundary markers: Absent. The skill instructions reinforce following the external specification without reservation.\n
- Capability inventory: The skill performs file system writes to create and modify TypeScript source code (Phase 5) and executes shell tools (
gh,tsc).\n - Sanitization: No sanitization or validation of the specification content for embedded commands or malicious logic is implemented.
Audit Metadata