The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Fetches documentation and guide PDFs from official Salesforce domains: developer.salesforce.com, help.salesforce.com, and resources.docs.salesforce.com. These are recognized well-known technology service domains, and the retrieval behavior is aligned with the skill's primary function.
[COMMAND_EXECUTION]: Uses the subprocess module to orchestrate internal helper scripts and launch a local Node.js scraper. This is used for legitimate task coordination within the skill's architecture and does not involve the execution of arbitrary commands from external inputs.
[PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection as it processes documentation retrieved from the web. Ingestion points: scripts/sync_sf_docs.py and scripts/retrieve_sf_docs.py. Boundary markers: None identified. Capability inventory: scripts/cli.py (subprocess calls) and scripts/sync_sf_docs.py (network access). Sanitization: Content is normalized via text extraction from HTML and PDF sources. The risk is considered acceptable given the skill's core purpose as a retrieval tool.
[SAFE]: No malicious patterns such as credential theft, persistence mechanisms, or obfuscated code were detected. The skill operates within its intended boundaries as a documentation grounding layer.

sf-docs