skills/ivanrvpereira/.agents/teach/Gen Agent Trust Hub

teach

Pass

Audited by Gen Agent Trust Hub on Sep 4, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONDYNAMIC_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ground its teaching in external data provided via RESOURCES.md and MISSION.md. It reads these files and learning records to determine lesson content and teaching strategy without defining explicit boundary markers or sanitization logic. This creates a surface where instructions embedded in retrieved articles or community forums could potentially influence the agent's behavior. Evidence includes the ingestion points: MISSION.md, RESOURCES.md, and ./learning-records/*.md.\n- [COMMAND_EXECUTION]: The skill explicitly instructs the agent to 'open the lesson file for the user by running a CLI command.' This capability allows for host-level interaction, which could be abused if the generated file paths are manipulated or if the agent is influenced to run unintended commands.\n- [DYNAMIC_EXECUTION]: The agent is tasked with generating interactive lessons and reusable components (such as quiz widgets and simulators) as HTML and JavaScript files in the ./assets/ and ./lessons/ directories. This involves the dynamic generation of executable code for browser contexts, derived from external and potentially untrusted knowledge sources.\n- [EXTERNAL_DOWNLOADS]: The skill's core function requires the agent to browse and identify high-quality knowledge resources from external websites and online communities to populate the workspace. This involves network activity and the processing of remote content that has not been pre-validated for safety.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 4, 2026, 02:31 PM