The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from external sources such as forum threads, social media posts, and GitHub issues to identify demand signals for early-stage customer discovery.
Ingestion points: Data is ingested from platforms like Reddit, Hacker News, X (Twitter), and YouTube using tools like last30days and social-fetch as described in references/demand-signals.md.
Boundary markers: The instructions include an explicit warning for the agent to treat fetched content as untrusted and to never follow instructions embedded within the processed content.
Capability inventory: The skill utilizes external APIs and CLI tools (Apollo, Clay, GitHub) and accesses local context files like product-marketing.md.
Sanitization: The skill recommends escaping all values when generating standalone HTML evidence reports from the gathered data.
[EXTERNAL_DOWNLOADS]: The skill references a centralized tools registry and various integration guides hosted on a public GitHub repository.
Evidence: SKILL.md and multiple reference files provide links to github.com/coreyhaines31/marketingskills for tool configurations and MCP documentation.
[COMMAND_EXECUTION]: The skill provides instructions for using a specific local command-line tool to perform technical prospecting.
Evidence: references/saas-prospecting.md contains a command for executing node tools/clis/github-prospects.js to extract and process data from GitHub repositories.

prospecting