The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from Apache Solr search results and text extracted from external files (PDF, DOCX, etc.) via Apache Tika. This constitutes an indirect prompt injection surface where instructions hidden in indexed data could influence agent behavior.
Ingestion points: Search result documents and metadata fetched from Solr (SKILL.md, references/07-search-configuration.md), and file content extracted via Tika (SKILL-SOLRFAL.md).
Boundary markers: None explicitly defined in the instructions to separate data from system prompts.
Capability inventory: Includes shell command execution via ddev exec and curl for debugging and API interaction (references/12-0-how-to-get-debug-data.md), file system modifications for site configuration, and network operations to Solr and external embedding APIs.
Sanitization: Employs TypoScript SOLR_CONTENT to strip HTML tags, which is primarily for content formatting rather than security isolation.
[COMMAND_EXECUTION]: The documentation provides extensive examples of shell commands that the agent is instructed to use for debugging, including ddev exec, curl, and solrctl (references/12-0-how-to-get-debug-data.md). While functional for the skill's purpose, these commands allow the agent to perform sensitive network and local system operations.

typo3-solr