The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The documentation in SKILL.md provides installation steps for local LLM support using Ollama, which involves executing a remote script via curl -fsSL https://ollama.com/install.sh | sh. Ollama is a recognized service for local model inference.
[EXTERNAL_DOWNLOADS]: The skill's primary functionality includes fetching content from arbitrary remote URLs for processing, as seen in the lx.extract() method and the scripts/langextract_extract.py CLI helper.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process unstructured data from external sources (files and URLs) which introduces a vulnerability to indirect prompt injection.
Ingestion points: The text_or_documents parameter in lx.extract() (documented in SKILL.md and references/api-reference.md) and the --input flag in scripts/langextract_extract.py.
Boundary markers: None explicitly identified in the provided instructions or scripts to separate data from instructions.
Capability inventory: The skill can perform network requests to fetch data, write results to the filesystem (lx.io.save_annotated_documents), and execute shell commands via the platform's allowed tools.
Sanitization: No specific input sanitization or filtering logic is present in the provided scripts or documentation.

langextract