ai-do
Pass
Audited by Gen Agent Trust Hub on May 7, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: Executes shell commands such as
ls skills/andls ~/.claude/skills/to discover which skill components are already installed in the user's environment. - [EXTERNAL_DOWNLOADS]: Fetches skill documentation, including
SKILL.md,examples.md, andreference.md, from thelebsral/DSPy-Programming-not-prompting-LMs-skillsGitHub repository to ensure the router has the latest information for its recommendations. - [PROMPT_INJECTION]: Acts as an indirect prompt injection surface by ingesting user-supplied descriptions of AI tasks to generate prompts for other skills. This risk is mitigated by a multi-step discovery process where the agent is instructed to ask clarifying questions before finalizing a routing command.
- [COMMAND_EXECUTION]: Recommends the use of the
npx skills addcommand to install missing components from the author's official repository as part of the routing workflow.
Audit Metadata