prospecting

Pass

Audited by Gen Agent Trust Hub on Oct 1, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to monitor and process user-generated content from external sources such as public forums (Reddit, Hacker News), social media posts, and product reviews to identify "pain signals" and early customer demand. This ingestion of untrusted external content represents a surface for indirect prompt injection, where an attacker could embed malicious instructions in a public post to influence the agent's behavior during discovery or outreach drafting.
  • Ingestion points: Browser-assisted research and specific tools like social-fetch, last30days, Firecrawl, and Browserbase are utilized to ingest content from the public web (SKILL.md, references/demand-signals.md).
  • Boundary markers: The skill employs structured qualification rubrics, pass/fail checklists, and Phase 1 ICP definitions to constrain how the agent interprets and uses ingested data (SKILL.md, references/demand-signals.md).
  • Capability inventory: The agent has capabilities for network communication via prospecting tool APIs, browser-based research automation, and the generation of local CSV files (SKILL.md, references/data-sources.md).
  • Sanitization: The instructions explicitly direct the agent to paraphrase signals and quote source material minimally, which acts as a foundational sanitization measure (references/demand-signals.md).
  • [COMMAND_EXECUTION]: The skill includes instructions for running a local Node.js utility, node tools/clis/github-prospects.js, to process GitHub repository data (stargazers and forks). This utility is used for lead enrichment and exporting data to structured CSV formats (references/saas-prospecting.md, references/data-sources.md).
  • [EXTERNAL_DOWNLOADS]: The skill references and integrates with a wide variety of well-known and established technology services and professional platforms, including Apollo, ZoomInfo, Clay, Clearbit, LinkedIn, GitHub, Crunchbase, Firecrawl, and Browserbase. These references are part of the skill's primary functionality for lead discovery and data enrichment (references/data-sources.md).
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 1, 2026, 02:19 AM