The Agent Skills Directory

[COMMAND_EXECUTION]: Utilizes curl to perform network requests for web scraping, probe for markdown-specific headers, and send mandatory status notifications to a local server at localhost:31337. It also uses echo to append execution logs to a local JSONL file at ~/.claude/LIFEOS/MEMORY/SKILLS/execution.jsonl.
[EXTERNAL_DOWNLOADS]: Interacts with api.brightdata.com, a well-known service provider, to trigger and monitor web crawling snapshots. It also downloads content from arbitrary URLs provided by the user across four scraping tiers.
[DATA_EXFILTRATION]: Accesses the BRIGHTDATA_API_KEY from a local .env file for API authentication and transmits execution metadata (workflow name, status, and duration) to a local notification endpoint.
[PROMPT_INJECTION]: The skill processes untrusted web content as its primary function, creating a surface for indirect prompt injection. Maliciously crafted data on scraped websites could attempt to influence the agent's behavior during the markdown conversion and output process.
[SAFE]: Includes a local customization feature that loads user preferences and resources from ~/.claude/LIFEOS/USER/CUSTOMIZATIONS/SKILLS/BrightData/. This is a standard and safe practice for providing local extensibility without relying on external or untrusted code sources.

BrightData