The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes an array of URLs from user input, which presents a surface for indirect prompt injection attacks if those URLs point to malicious content that the agent then processes.
Ingestion points: The urls parameter in the scrapeMany function (SKILL.md, Step 4) and the bulkScrape function (SKILL.md, Step 5).
Boundary markers: None present; the code does not use delimiters or instructions to ignore embedded commands in the fetched content.
Capability inventory: The skill uses axios and fetch to perform network operations, and has permission to Read, Write, and Edit files.
Sanitization: No sanitization or validation of the input URLs is performed before they are used in network requests.
[DATA_EXFILTRATION]: The skill's connection pooling implementation (Step 2) includes rejectUnauthorized: false in the HTTPS agent configuration. This disables SSL certificate verification for connections to the brd.superproxy.io proxy endpoint, creating a vulnerability where data retrieved or proxy credentials could be intercepted via Man-in-the-Middle (MitM) attacks.

brightdata-performance-tuning