agent-asset-studio
Pass
Audited by Gen Agent Trust Hub on Jul 22, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill invokes bundled Python scripts (
init_asset.py,validate_asset.py,routerize_skills.py,update_index.py) to automate asset management tasks. These scripts perform filesystem operations such as creating directories, writing files, moving resources, and modifying CSV indexes. - [COMMAND_EXECUTION]: The
scriptifyworkflow (workflows/scriptify.md) and its supporting references (scriptify-apply.md) facilitate the conversion of natural language procedures into executable scripts. This process involves the agent writing code to a file and subsequently executing it, which is an inherent capability for its optimization goals. - [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface because it is designed to ingest and evaluate existing Markdown assets (e.g.,
AGENTS.md, skill files, and rules) that may originate from potentially untrusted sources within a repository. Although the skill instructs the agent to treat these contents as data, the capability to read and then act upon the content of these files (e.g., during 'evaluation' or 'improvement' workflows) exists. - [SAFE]: The skill references the official Agent Skills documentation and specification at
agentskills.iofor informational purposes, which is a trusted domain for this context.
Audit Metadata