The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill scripts scripts/extract-extension-config.sh and scripts/check-ui-surface-screenshots.sh use the PHP include directive to load and execute local project configuration files (ext_emconf.php and Configuration/Backend/Modules.php) during the metadata extraction process. This represents dynamic loading of code from computed paths within the project being documented.\n- [EXTERNAL_DOWNLOADS]: The skill fetches the ghcr.io/typo3-documentation/render-guides:latest Docker image to render documentation and utilizes official GitHub and GitLab APIs to retrieve repository metadata. These downloads target well-known and reputable services and are documented neutrally in accordance with safety guidelines.\n- [INDIRECT_PROMPT_INJECTION]: As a documentation tool, the skill processes untrusted extension data including PHP source code, TypoScript configurations, and TCA definitions. This creates an attack surface for indirect prompt injection.\n
Ingestion points: The skill reads extension source code and configuration files from the local project directory (e.g., in scripts/extract-php.sh and scripts/analyze-docs.sh).\n
Boundary markers: No specific delimiters or boundary markers are mentioned in the instructions for processing these files.\n
Capability inventory: The skill possesses shell command execution capabilities (via tools like php, docker, sed, grep) and file writing capabilities (to Documentation/ folders).\n
Sanitization: No explicit sanitization or filtering of external content is described before it is processed by the agent or interpolated into documentation templates.\n- [COMMAND_EXECUTION]: The skill executes various system commands and tools, including php, docker, sed, grep, and find, to perform analysis, validation, and rendering of TYPO3 documentation.

typo3-docs