learning-loop
Pass
Audited by Gen Agent Trust Hub on Jul 12, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [SAFE]: The skill is composed of instructional Markdown files and templates intended to guide an AI mentor's behavior. It does not contain any executable scripts, shell commands, or network-facing operations.
- [NO_CODE]: There are no Python or Node.js packages required for this skill. It functions entirely through prompt instructions and local file management.
- [PROMPT_INJECTION]: Analysis of the indirect prompt injection surface:
- Ingestion points: The skill reads content from the user's local repositories to design and review learning exercises (SKILL.md).
- Boundary markers: The skill does not define specific delimiters to separate ingested repository data from the agent's internal instructions.
- Capability inventory: Operations are limited to reading repository content and writing local Markdown artifacts in the .ai/learning/ directory.
- Sanitization: No explicit validation or filtering of the ingested repository content is mentioned.
- Assessment: While a surface for indirect prompt injection exists via the user's repositories, the risk is negligible as the skill lacks capabilities for network communication or arbitrary command execution.
Audit Metadata