MCP Tools Reference

​

Overview

When you connect OpenFunnel to Claude via MCP, a set of tools becomes available, and Claude will automatically pick the right one based on your question.

​

Two families: data fetchers vs. data generators

OpenFunnel MCP exposes two kinds of tools:

• Data fetchers read what’s already in your workspace — accounts, audiences, insights, signals. They are free and have no rate-limited credit cost. Most tools below fall in this group.
• Data generators are wrappers over the Agent Primitive APIs. They generate new data on demand — searching the broader company universe, surfacing companies hiring for a given technology, etc. They consume credits per company returned and are clearly marked with a credit warning. Claude will confirm with you before calling them.

Credit-consuming generator tools currently:

• search_lookalikes — 1 credit / company
• search_companies_by_tech_stack — 5 credits / company unenriched, 10 credits / company enriched
• deep_research — 1 charge per call (per-call price set by your workspace; echoed back in credits_consumed)

---

​

get_account_summary

Get a quick summary/overview of an account. Parameters: Returns: Basic account info, signal counts, people count, and TOP 3 key contacts only.

---

​

get_account_full_details

Get COMPLETE account details including ALL people/contacts at the company. Example queries:

• Who works at [company]?
• Show me all contacts at [company]
• Get me people at [company]
• List employees at [company]
• Who are the decision makers at [company]?
• Any request for people/contacts/employees at a company

Parameters: Returns: Complete account data with ALL people and ALL signals.

---

​

get_account_timeline

Get recent activity timeline for an account. Parameters: Returns: Timeline with chronological list of events.

---

​

get_my_insights

Find out what has been happening with your accounts in the last n days. Example queries:

• Show me what has been happening with my accounts in last n days?
• Show me net new accounts in last n days?
• Tell me more about the alerts that were sent to me today?

Parameters: Returns: Alerts with total_alerts, has_more, and alerts list (each with embedded insights). If no alerts exist, returns fallback insights_feed data instead.

---

​

get_insight_detail

Get full detail for a single insight. Drill down from the feed. Example queries:

• Tell me more about this insight
• What signal triggered this alert?
• Show me the details for insight X
• Any drill-down from the insights feed

Parameters: Returns: Insight details with underlying_signal and related_insights.

---

​

list_accounts

List accounts in your database with optional filters. Parameters: Returns: List of accounts with pagination info.

---

​

list_audiences

List all audiences (saved lists) for the user. Audiences can be either “accounts” type (primary entities are accounts) or “people” type (primary entities are people/prospects). Use this to discover what audiences exist before fetching their data with get_audience_data. Returns: List of audiences with audience_id, audience_name, and audience_type.

---

​

get_audience_data

Get all members on an audience. Use this to retrieve the members of an audience. Works for both “accounts” and “people” audience types. Parameters: Returns: Audience metadata plus member IDs and totals for accounts and people.

---

​

lookup_company

Look up a specific company in the user’s OpenFunnel database by name or domain. Example queries:

• Tell me about [company]
• What’s happening with [company]
• Show me [company] from OpenFunnel
• Find [company] in my accounts
• Any question about a specific company by name or domain

Parameters: Returns: Company summary including basic info, signal counts, and key contacts. If multiple companies match, returns options to clarify which one.

---

​

search_by_traits

Search within your existing database for companies matching specific characteristics. Use this to find companies in your DB matching things like:

• Industry or vertical (healthcare, fintech, etc.)
• Products or services they offer
• Technologies they use
• Business model or target market

Parameters: Returns: List of matching companies in your database, ranked by relevance.

---

​

search_lookalikes

Search the broader company universe (beyond your database) for net-new prospects, with optional filters for size, funding stage, and location. Use this when you want to find companies you don’t already have, e.g. finding lookalikes of your best customers, or expanding into new geographies / segments.

​

Seed-based discovery (no need to phrase the trait)

In addition to natural-language query, you can pass up to 10 company websites in seed_domains and OpenFunnel will use them to synthesize the lookalike query for you. Useful when you can name a few example customers but struggle to phrase what unifies them.

• Seeds only: pass seed_domains — the API picks the common theme automatically and surfaces it in the derived_query field of the response.
• Seeds + query (must-have angle): pass both — the query becomes a constraint on top of the trait the seeds imply (e.g. seed_domains=["stripe.com", "adyen.com"] + query="for SMB merchants" → synthesized trait targets B2B SMB payments platforms).

Unresolved seeds (e.g. typos, very small companies not in our index) are skipped silently as long as at least one seed resolves; the response reports both resolved_seed_domains and unresolved_seed_domains so you can show the user exactly what landed. At least one of query or seed_domains is required.

​

Choosing a search strategy: precision vs. recall

search_lookalikes supports two retrieval strategies via the search_type parameter: Rule of thumb: start with agentic if the query has a clear category (e.g. “B2B customer support software”, “LLM evaluation tools”, “HVAC home services”); switch to semantic if results feel too narrow or the trait is fuzzy. Parameters: Returns: Object with query, results (each entry: name, domain, linkedin_url, headquarters, company_size, match_reason), total, credits_consumed (1 per returned company), and — when seed_domains was used — derived_query (the LLM-synthesized trait), resolved_seed_domains, and unresolved_seed_domains.

---

​

search_companies_by_tech_stack

Find companies whose recent job postings mention a given technology, ranked by matching post count. Use this for technology-adoption discovery — e.g. “Find companies hiring for Snowflake”, “Show me US-HQ companies of 1,000+ employees adopting Kubernetes”, “Series B+ fintechs hiring for ClickHouse engineers”. Page size is fixed at 50; walk pages with page=1, page=2, … using has_more from the response. Parameters: Returns: A page of matched companies with their top-hit job snippet (when include_evidence=true) and firmographics (when enriched), plus has_more for pagination and credits_consumed for the exact amount billed.

---

​

deep_research

Decide whether a single, known company fits your criteria in terms of recent activity and organizational structure as a qualifier. Combine an activity question (answered from job postings) and a team-composition question (answered from LinkedIn people) in a single call, with cited evidence per stage. Provide a domain plus at least one of activity_question or qualifier_question. Each question is routed to a specific data source — pick the param accordingly:

• activity_question → answered from recent job postings (use for hiring / team-expansion / JD-visible questions).
• qualifier_question → answered from LinkedIn people data (use for team composition / org-shape questions).

Use this to research an account before outreach — e.g. “Is Stripe hiring SDRs?” (activity_question) + “Do they already have a Sales Director?” (qualifier_question), or “Does Notion have a dedicated AppSec team?” (qualifier_question). Pass both when the question spans both signals.

​

Two stages, two data sources

Each parameter is hard-wired to a specific data source. Pick the param that matches where the answer lives: You can pass either or both — at least one is required. Pass both when a question spans both signals (e.g. “they have an SDR team and are hiring more SDRs”) — each stage runs against its own data source and returns independent evidence.

​

When to use this tool vs. others

• Use deep_research to research a known company (you have the domain) and answer specific yes/no research questions with evidence.
• Use search_lookalikes when you don’t yet have the company — net-new discovery across the broader universe.
• Use search_companies_by_tech_stack for technology-adoption discovery across many companies.

Parameters: Returns: An object with:

• company — resolved firmographics for the domain (name, employee count, stage, industry, HQ location, short description, LinkedIn slug)
• activity — null when not asked, otherwise { qualified, answer, confidence, jobs_searched, jobs_matched, sources: [...job postings], elapsed_ms }
• qualifier — null when not asked, otherwise { qualified, answer, detailed_reasoning, people_evaluated, sources: [...profiles], elapsed_ms }
• credits_consumed — exact amount billed for this call (0 on a 404)
• latency_ms — end-to-end request latency