The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes source code files from the user's project, which creates a surface for indirect prompt injection if those files contain adversarial instructions in comments or metadata.
Ingestion points: The skill reads PHP classes, Fluid templates, TCA configuration files, YAML services, and TypoScript files (SKILL.md, sections 1-10).
Boundary markers: No explicit delimiters or instructions to ignore embedded instructions are defined for the processed content.
Capability inventory: The agent is instructed to modify files, execute shell commands (php -l, composer normalize, vendor/bin/phpunit), and perform git commit operations (SKILL.md, Execution Rules).
Sanitization: No explicit sanitization or filtering of the ingested code content is mentioned.
[COMMAND_EXECUTION]: The skill directs the agent to execute several shell commands for validation purposes.
Evidence: The instructions include running composer normalize, php -l, and vendor/bin/phpunit after performing code changes (SKILL.md, Execution Rules and Verification Checklist). These are standard development tools used for linting and unit testing within the TYPO3 ecosystem.
[EXTERNAL_DOWNLOADS]: The skill includes references to external documentation and source code repositories.
Evidence: It links to official TYPO3 documentation (docs.typo3.org) for event mapping and specific breaking changes, as well as the author's public GitHub repository (github.com/dirnbauer/webconsulting-skills). These are recognized as legitimate and relevant resources for the skill's stated purpose.

typo3-batch