The suite is read-only apart from push_leads_to_sequencer, which adds leads to an Instantly or Smartlead campaign — and even that takes dry_run, a minimum ICP score and deduplication against what the campaign already holds, so the first pass can be a dress rehearsal. The output shape is the other deliberate decision: flat, one row per company, which is what lets these calls drop straight into an enrichment table instead of being reshaped first. Cost sits on the Apify side, one actor run per call against your token, so point aggregate_gtm_signals or get_company_changes at a shortlist rather than a whole list.
A single MCP server over the Mamba Labs account-intelligence actors on Apify. Each tool wraps one actor and answers in flat, Clay-ready JSON, covering a full go-to-market pass: resolve the identity, enrich the account, detect buying signals, score fit, then send the survivors on. The package is a thin client — the work runs on Apify against your own token.
- resolve_company_identity reconciles a name, a domain and a LinkedIn URL into one canonical identity, each part scored 0-100 with the match method attached; resolve_linkedin_url does the narrower job alone
- enrich_company_firmographics reads employee band, industry, HQ, founded year, revenue estimate, logo and description from a site's schema.org/Organization JSON-LD and meta tags, with source provenance
- map_company_social_presence finds LinkedIn, X, Instagram, Facebook and YouTube profiles with follower counts, validating what it finds rather than pattern-guessing and hoping
- Signals: scan_gtm_hiring_signals across Greenhouse, Lever and Ashby, scan_job_board_keywords adding Workday and Rippling against your own role categories, detect_gtm_tech_stack for CRM, sequencer and marketing automation, and get_funding_press_signals reading Google News and PR wires for funding, executive moves, launches and acquisitions
- get_company_changes monitors a domain and returns only what moved since the last run, as typed change events
- detect_ai_tooling grades a company none, declared, deployed or commercialized — the last meaning the pricing page actually charges for AI rather than the homepage mentioning it
- fingerprint_outbound_infrastructure answers whether a company runs cold outbound — program, light, none or unknown — with the lookalike sending domains as its strongest evidence
- track_publication_cadence counts long-form posts over 30 days, 90 days and 12 months and calls the trend accelerating, steady, declining, dormant or unknown
- score_icp_fit returns a 0 to 100 score, an A to D tier and a per-signal breakdown, defined by a template, a scoring_config or a plain-English icp_description; aggregate_gtm_signals runs hiring and tech stack together for a composite score and a recommended action
- push_leads_to_sequencer is the one tool that writes: it maps Clay column names onto Instantly's or Smartlead's own fields, can drop leads below a minimum ICP score, deduplicates against the campaign, and honours dry_run
Node.js 18 or newer, run over stdio as npx -y @mambalabsdev/mcp-gtm-suite. APIFY_TOKEN is required — from console.apify.com/account/integrations — and every call runs an Apify actor and consumes Apify credits. score_icp_fit needs an llm_api_key only when the ICP is given in plain English rather than as a template or scoring_config. push_leads_to_sequencer additionally takes the sequencer's own api_key and a campaign_id. MIT.
One command plus a key — npx -y @mambalabsdev/mcp-gtm-suite, then supply credentials