zensical
Pass
Audited by Gen Agent Trust Hub on Nov 1, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill requires the execution of shell commands to install dependencies and run the documentation tool's build and preview servers. Evidence includes commands like '.venv/bin/python -m pip install "zensical==0.0.74"' and 'zensical build --strict'.\n- [EXTERNAL_DOWNLOADS]: The skill facilitates the download and installation of the 'zensical' package and its dependencies from public registries. It also references external documentation and resources from the 'zensical.org' domain and official GitHub repositories.\n- [INDIRECT_PROMPT_INJECTION]: The skill identifies a vulnerability surface where the agent might process untrusted documentation files containing malicious macros or Python execution fences.\n
- Ingestion points: Project documentation files and configuration files (e.g., 'zensical.toml', 'docs/*.md').\n
- Boundary markers: The skill includes multiple warnings in 'references/configuration.md' and 'references/publishing.md' advising the user to treat macros as executable input and to avoid building untrusted content when secrets are accessible.\n
- Capability inventory: The documentation builder ('zensical build') has the capability to execute embedded Python code and macros via extensions like 'markdown-exec'.\n
- Sanitization: No specific sanitization or filtering of macro content is implemented in the provided instructions.\n- [DYNAMIC_EXECUTION]: The skill documents and enables the use of dynamic execution features within the documentation generator, such as Python Markdown extensions and Jinja-based macros, which can execute code at build time.
Audit Metadata