The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the official bdata CLI and curl to interact with the Bright Data Discover API. These commands are documented for their intended use in search and retrieval workflows.
[DATA_EXFILTRATION]: Network communication is directed to api.brightdata.com, which is the authoritative domain for the vendor's API. No unauthorized data transmission was detected.
[PROMPT_INJECTION]: The skill retrieves untrusted third-party data from the web (page content, titles, and descriptions), creating an indirect prompt injection surface.
Ingestion points: The results[].content, results[].title, and results[].description fields (SKILL.md) ingest external content.
Boundary markers: No explicit delimiters or instruction-override warnings are provided for the agent when processing these fields.
Capability inventory: The skill has access to the shell via curl and the bdata CLI (SKILL.md).
Sanitization: The documentation suggests filtering for block-page and CAPTCHA signatures, but does not include specific sanitization for malicious instructions.

discover-api