The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill provides patterns for parsing HTML from external websites, which serves as an ingestion point for untrusted data into the agent's context.
Ingestion points: The parseProductPage(html) and parseSearchResults(html) functions in SKILL.md accept arbitrary HTML content from external sources.
Boundary markers: Absent; the parsed content is returned directly as structured objects without delimiters or instructions for the agent to ignore embedded commands.
Capability inventory: The skill is configured with Read, Write, and Edit tools, allowing the agent to perform file system operations based on the data it parses from untrusted sources.
Sanitization: Absent; the HTML is parsed using cheerio but the resulting text is not sanitized to remove or escape potential prompt injection attempts.
[COMMAND_EXECUTION]: The TypeScript proxy client implementation in SKILL.md explicitly disables SSL/TLS certificate validation, which is a significant security misconfiguration.
Evidence: The code snippet for getBrightDataClient includes httpsAgent: new https.Agent({ rejectUnauthorized: false }).
Impact: This configuration makes all proxy traffic vulnerable to Man-in-the-Middle (MitM) attacks, allowing attackers to intercept or modify the scraped data despite the use of HTTPS.

brightdata-sdk-patterns