subagent-prompt
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides structured guidance for the AI to create a self-contained prompt for a secondary session. It emphasizes logical decomposition and verification of work rather than executing dangerous actions.
- [INDIRECT_PROMPT_INJECTION]: The skill identifies and reads external artifacts such as plan files (
plan.md), checklists, and project manifests (package.json,Cargo.toml,Makefile). While this represents a surface where untrusted data could be ingested, the skill's primary output is a text-based prompt for the user to copy and paste. The risk is minimized by the instructional gates requiring the agent to point to concrete artifacts and specific commands rather than blindly following embedded instructions. - [COMMAND_EXECUTION]: The skill instructs the agent to search for and include project-specific commands (e.g.,
cargo test) in the generated output. This is a legitimate development use case and the skill itself does not invoke these commands.
Audit Metadata