ai-coding-agents
Pass
Audited by Gen Agent Trust Hub on Aug 12, 2026
Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill provides templates for agents (e.g., in
assets/templates/code-reviewer.md) that ingest and process untrusted repository content, creating an inherent surface for indirect prompt injection. This risk is addressed through explicit guidance inreferences/debugging-guide.md, which instructs agents to treat code as data and ignore embedded directives. - [EXTERNAL_DOWNLOADS]: The
scripts/smoke_test.shutility performs environment validation by checking the reachability of the official Anthropic and OpenAI APIs. These requests target well-known services and do not involve sensitive data harvesting. - [COMMAND_EXECUTION]: Python and TypeScript templates provided in
assets/templates/(e.g.,sdk-agent-py.py) contain standard code for executing shell commands via subprocess and child_process modules. This capability is required for the intended purpose of wrapping development tools like linters and test runners for agent use.
Audit Metadata