Skip to content

discolike/discolike-skills / skills/find-validated-contacts

find-validated-contacts

Verified

Use when the user asks to "find VP Sales contacts at these companies", "get validated work emails", or deliver contacts at scale. Follow the discolike skill's access and spend rules.

Deliver validated contacts

Follow the discolike skill's How to work rules first.

Load the discover-lookalike-companies skill if you need a target list and the check-icp-fit skill if you need to check company fit first.

Use MCP for contacts.

  1. Start contacts. MCP find-contacts takes a run object. Select companies using one of the objects below. Set run.people.fit to a plain-language description of the people wanted; the default is "Senior executives and decision makers". Set run.max_companies (1 to 10,000, default 500) and run.per_company (1 to 50, default 10). max_companies counts companies delivering at least one validated contact. Optional run.email_cap limits delivery spend in USD; run.create_key makes a start retry idempotent. run.exclude_query_ids and run.exclude_domains exclude companies already worked. The run starts automatically.

    Company selections for run.companies:

    • Prompt: {"source": "prompt", "icp_text": "US cybersecurity consultancies"}.
    • Domain list: {"source": "domains", "domains": ["example.com"]} (up to 10,000 domains).
    • Lookalikes: {"source": "lookalike", "persona_ids": [123]} using actual persona IDs from prior results, or {"source": "lookalike", "linkedin_urls": ["https://www.linkedin.com/in/example"]} with up to 20 LinkedIn profile URLs of people like the ones wanted (both may be combined). Each profile is resolved to its current title and company; a profile with no current position or no matching company stays unresolved, and a run whose seeds all fail to resolve is rejected with a 422.

    Example MCP arguments:

    {
      "run": {
        "companies": { "source": "domains", "domains": ["example.com"] },
        "people": { "fit": "Senior executives and decision makers" },
        "max_companies": 1,
        "per_company": 3,
        "email_cap": 1
      }
    }
    
  2. Follow and page results. Poll get-contacts-run with run_id. Pass the returned next_cursor as events_after for new progress events. rows_offset and rows_limit page companies, keeping each company's contacts together; MCP allows 1 to 100 companies per page (default 50). Advance the offset by the page size and compare with companies_total; rows_total counts contacts, not pages. companies contains live company status and contacts; rows contains the delivered contact rows. Optional fields selects fields in rows. Omit run_id to list 20 recent runs and use before to page older ones. Runs expire after 30 days idle.

  3. Read validated delivery. Contacts appear only with status Validated or Validated (catch-all resolved) and a validated work email. Do not re-verify either status. Validated (catch-all resolved) means the mailbox was proved on a domain that accepts all mail; another verifier may report catch-all or unknown. Every delivery, including repeats, costs 5 records per validated contact with a $10 per 1,000 contacts floor, plus 1 record per company delivering at least one when searching without a supplied company list. Companies with none are free. On completion or cancellation, report summary.companies_delivered, summary.validated_contacts and summary.spent_usd. Report shortfalls rather than assuming the requested target was met.

    • LinkedIn contacts (opt-in). Set run.include_linkedin_only: true only when the user wants people whose work email could not be verified. They are delivered without an email, with their LinkedIn profile URL and status LinkedIn, at 1 record each instead of the contact delivery price. They count toward run.max_companies and run.per_company, are left out of push-to-crm, which sends only validated email contacts, and must be presented as LinkedIn-only, not as emailable leads. Agree this with the user before starting, since it changes what the run delivers and costs.
  4. Stop or deliver. Use cancel-contacts-run with run_id to stop and retain already validated results; reading and cancellation remain available after quota exhaustion. Export validated results as CSV, or use push-to-crm with run_id and provider (hubspot, salesforce, pipedrive). It accepts an owned run with completed or partial validated results. Explain when results are partial. Supply exactly one of run_id or query_id; mode="enrich" updates existing CRM records only, while the default mode="push" creates or updates them.

Contacts instructions here describe the verified MCP contract. Check the installed CLI/SDK help before using equivalents; do not translate these arguments into guessed commands. The released POST /contacts/discover contract is unchanged.

Batch large contact lists

For contacts, use the same validated delivery flow at every size (steps 1 to 4 above). Split company domains into batches of at most 10,000 and set run.max_companies, run.per_company and an agreed run.email_cap for each batch. Poll and page each run by company. Export rows as they arrive and record the run ID and progress cursor so recovery reads the existing run. Use run.create_key when retrying the same start request. Starting a new run delivers and bills contacts again, even for companies already processed. Do not estimate yield or throughput from the maximum per company; report actual validated contacts and companies covered.

reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 471kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 322kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 322kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 322kUpdated Claude CodeCodexVerifiedInstall