Open-Web Prospect Research

skill

Search, navigate, and extract signals on companies and people from the open web to fuel prospecting.

Download .md

Open-Web Prospect Research

Search, navigate, and extract signals on companies and people from the open web to build and enrich prospecting lists.

When to use

  • A task asks you to find or qualify prospects, companies, or contacts not already in a CRM.
  • You need firmographic, funding, hiring, or tech-stack signals to prioritize outreach.
  • You must verify or refresh stale facts about a target account before a campaign.
  • A delegated task hands you a list of company or person names needing research.

Tools

  • browser-search: runs web searches to find candidate pages, news, and profiles for a target company or person.
  • browser-navigate: opens a specific search result or known URL to load the page content.
  • browser-extract: pulls structured signals (name, title, company, funding, headcount, tech stack) out of a loaded page.

Playbook

  1. Clarify the target: company name, person name, or ICP (ideal customer profile) criteria given in the task.
  2. Run browser-search with the target name plus qualifying terms (for example, "company name" + "funding" or "company name" + "careers").
  3. Review the search results and pick the most authoritative or recent source (official site, press release, LinkedIn-indexed page, news outlet).
  4. Use browser-navigate to open that result.
  5. Use browser-extract to pull the specific signal you need (title, headcount, tech stack, recent news, contact role).
  6. If the extracted page lacks the signal, return to browser-search with refined terms and repeat steps 3-5.
  7. Repeat the search-navigate-extract cycle for each additional signal or each prospect in the list.
  8. Compile findings into a structured note (company, signal, source URL, date found) for hand-off or storage via write-file.
  9. Summarize the research and its confidence level when completing the task.

Failure modes

  • 401/403 on any page load -> not applicable to open web browsing; if a source requires login, skip it and note the gap rather than attempting credentials.
  • Empty or irrelevant search results -> refine query terms (add location, industry, or alternate name spellings) and retry browser-search.
  • Rate limiting or blocked navigation -> back off, try an alternate source, and note the limitation in memory via update-memory.
  • Target page structure changed or extraction returns nothing usable -> fall back to browser-search for an alternate source covering the same signal.

Done when

  • Each requested signal has a value, a source URL, and a retrieval date, or is explicitly marked as unavailable.
  • The compiled research note is ready to hand off (via write-file/complete-task) with no unverified guesses presented as facts.
  • Any blocked or inconclusive lookups are called out rather than silently omitted.