The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill makes requests to api.firecrawl.dev to utilize its specialized scraping and branding extraction service. This is a core feature of the skill, and the domain is the official endpoint for the Firecrawl service.
[EXTERNAL_DOWNLOADS]: Both basic-extract.sh and firecrawl-extract.sh use curl to fetch content from user-provided target URLs and official service endpoints.
[EXTERNAL_DOWNLOADS]: The skill fetches visual assets (logos, screenshots) and references fonts.googleapis.com for font extraction, which are standard operations for building a brand profile from a live website.
[COMMAND_EXECUTION]: The skill utilizes shell scripts (scripts/basic-extract.sh, scripts/firecrawl-extract.sh) to perform data processing. These scripts use common utilities like grep, sed, jq, and python3 to parse and structure the extracted data.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external websites (HTML, text content, metadata) into the workspace as extract.md and extract.json.
Ingestion points: Raw HTML/markdown fetched from user-provided URLs in both basic-extract.sh and firecrawl-extract.sh.
Boundary markers: The skill uses a structured extraction process and explicit instructions to "Preserve exact language where possible. Quote, do not paraphrase." and "Distinguish observed facts from inference," which helps segregate untrusted data from instructions.
Capability inventory: File system writes (workspace directory), network operations via curl (scraping), and shell command execution for data parsing.
Sanitization: The scripts perform basic data cleaning via sed and grep and use json.dumps (Python) or jq to safely format data into JSON output files.

site-extract