teaching-suite

Warn

Audited by Gen Agent Trust Hub on Aug 29, 2026

Risk Level: MEDIUMDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [DYNAMIC_EXECUTION]: The lab-forge workflow utilizes the solution_verifier_agent to execute code provided as student-facing materials to confirm the functionality of lab packages.
  • This process involves running potentially untrusted code from student assignments to verify reference solutions in the lab-forge/agents/solution_verifier_agent.md component.
  • [DYNAMIC_EXECUTION]: The deck-studio workflow generates and executes Python and LaTeX scripts via the figure_maker_agent and renderer_agent to produce diagrams and slide decks.
  • It uses libraries like matplotlib and python-pptx and executes generated code to render visual assets based on user-provided outlines.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes a significant amount of external, untrusted data, which presents an attack surface where malicious instructions could attempt to subvert the agent's behavior.
  • Ingestion points: The skill ingests student submissions for auditing (ts/submission-auditor/), student evaluation comments for analysis (ts/teaching-reflector/), and accreditation standards provided by the user (ts/accreditation-mapper/).
  • Boundary markers: The skill employs specific templates and instructions, such as the ts/shared/checkpoint_protocol.md, that require the agent to surface evidence and request manual verification through [VERIFY] and [NEEDS PROFESSOR INPUT] markers.
  • Capability inventory: The skill is configured to search the web, read and write local files, and execute shell and Python commands through the Bash tool.
  • Sanitization: The provided scripts use standard libraries for safe parsing with yaml.safe_load and output escaping with html.escape for generating HTML dashboards.
  • [COMMAND_EXECUTION]: The suite is designed to interact with and execute various command-line tools.
  • CLI tools such as marp, pandoc, and tectonic are used to facilitate the rendering of teaching materials as specified in ts/deck-studio/references/toolchain_guide.md.
  • [EXTERNAL_DOWNLOADS]: Instructions within the deck-studio reference files guide the user to download and install third-party software.
  • This includes external rendering tools and libraries required for the suite's production features, documented in ts/deck-studio/references/toolchain_guide.md.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Aug 29, 2026, 03:39 PM
Security Audit — agent-trust-hub — teaching-suite