skillZs
★ LIVE SKILL TAGS ★
>>> LIVE SKILLS INDEX <<<
* OPEN SOURCE *
NO LOGIN, NO TRACKING
※ REAL INSTALL DATA ※
← back to all skills
docs.clickmax.io121 installs

clickmax-leads

Use when the user wants to create, find, inspect, filter, verify the e-mail of, or compare CRM leads and their commercial context inside Clickmax.

How do I install this agent skill?

npx skills add https://docs.clickmax.io --skill clickmax-leads
view source ↗

Is this agent skill safe to install?

No partner audit is available yet. Read the source before installing.

What does this agent skill do?

As tools abaixo aparecem com os nomes que o MCP da Clickmax registra. Se o seu cliente de IA prefixar nomes de tool (mcp__<servidor>__, mcp_<servidor>_, ou outro), use o nome já prefixado que aparecer na sua lista de tools.

When this applies

Use this skill when the user wants lead discovery/inspection OR to create a lead: search/filter contacts, inspect one lead, check whether an email already exists, verify a contact's e-mail, find contacts with e-mail problems or fraud signals, create a new contact, compare lead-origin patterns, or pull a lead's payments/invoices/products.

Not this skill:

  • tagging/classification -> clickmax-tags
  • manual lists or dynamic segments -> clickmax-list-segments
  • kanban pipelines/opportunity cards -> clickmax-pipelines

Key assumptions

  • leads_filter_schema describes the legacy filter model; never use it to build leads_search filters
  • leads_search is the main entry for cohort discovery. filters is required ([] = all leads), and perPage accepts 1-100. Its only sortBy values are opportunities_asc/opportunities_desc; omitting sortBy returns leads NEWEST-FIRST by createdAt. So "the last N leads" uses filters: [], page: 1, perPage: N (N <= 100) with no sortBy
  • leads_get is enriched commercial context, not just a flat row
  • payments, invoices, common products, and origin trees are lead-adjacent projections, not separate core entities
  • tags, lists, and segments group leads; they do not replace the lead record itself
  • lifecycle and temperature are mutable business signals; report them as current state, not immutable history
  • e-mail status lives on the lead (emailStatus, emailValidatedAt, emailInvalidReason) and comes back in BOTH leads_search rows and leads_get: pending = not verified yet | valid | invalid (bounces; fix or exclude) | risky (deliverable but disposable/catch-all/role address; sending is still allowed, the user decides). Reading it costs nothing — do not verify just to read it
  • valid from the automatic background pass only confirms the DOMAIN; leads_validate_email asks for a mailbox-level check. Editing a contact e-mail already resets it to pending and triggers a check by itself
  • leads_validate_email is a PAID provider lookup (1 credit per address), synchronous, limited to 10 calls/min, and repeating it for the same lead within 1 hour just returns the stored verdict. It is for one specific contact, never a base-wide sweep
  • suspectedFraud (+ suspectedFraudReasons email/document) is a read-time warning of card testing, true only when BOTH the e-mail has no relation to the name AND the document is suspect (empty, all zeros, or not 11/14 digits). Nothing is stored or blocked; fixing the e-mail or document clears it. It is a filter field like any other, so it also works in segments

Thought process

  1. Decide whether the user needs one lead, a filtered cohort, or supporting aggregates.
  2. If the user describes filters vaguely, inspect the filter schema first.
  3. Use leads_search for cohorts and leads_get for one concrete lead.
  4. Pull supporting projections only when they materially answer the request.

Execute guide

  • For cohort discovery, follow the leads_search input contract; do not translate fields from the legacy filter-schema operation.
  • Search cohorts with leads_search, passing the required filters array ([] when unfiltered) plus optional paging and sort fields. Use this for discovery, comparison, and broad CRM filtering.
  • Inspect one known lead with leads_get, passing the lead id. Treat this as the main enriched lead view.
  • Add commercial context with leads_payments, leads_invoices, and leads_common_products only when payments, billing status, or bought-product patterns materially change the answer. Unlike its siblings, leads_common_products requires filter (not optional) — always pass a filter, even a broad one.
  • Use leads_exists_by_email for duplicate-check questions, not enrichment.
  • E-mail health of a cohort: filter leads_search with a filter item on field emailStatus, operator equals, valueString = invalid | risky | pending | valid (add negation: true for "anything but"). Fraud signals: a filter item on field suspectedFraud, operator equals, valueBool = true. Report the count (meta.countItens) before rows.
  • Verify one contact's e-mail with leads_validate_email, passing the lead id: it returns the new emailStatus right away. A pending result after a failure (provider error / inconclusive) is not a verdict — the automatic job retries later; do not retry in a loop. A lead without e-mail cannot be verified.
  • "How many contacts do I have in total?" -> analytics_resource_counts (contacts, merged contacts excluded) is one cheap call; use leads_search meta.countItens when the count must respect filters.
  • Create one contact with leads_create — only name is required; pass email/telephone when known and check leads_exists_by_email first to avoid duplicates. It also accepts tagIds/customFieldValues inline, so a lead can be created pre-tagged/pre-classified in the SAME call instead of a separate tagging step afterward. It returns the new lead id. To seed a pipeline, create each contact here then add them as opportunity cards via clickmax-pipelines (cards_create needs the returned lead ids). For several contacts, call leads_create once per contact.
  • Use leads_origins, leads_sub_origins, and leads_origins_tree for source taxonomy and breakdown questions.
  • Use leads_payments_utm_autocomplete when the user needs help discovering UTM values before filtering or diagnosing acquisition patterns.
  • Preferred order: cohort question -> leads_search; single lead question -> leads_get -> supporting projections only if needed; origin or UTM exploration -> origin or UTM helper first -> lead search only when matching contacts are also required.

Report

  • Start with what was inspected: one lead, cohort, or origin/UTM diagnostic.
  • For one lead: summarize identity, status/context, and only the relevant commercial facts.
  • For cohorts: summarize count + the most relevant breakdowns before dumping rows.
  • Cap long result sets and show +N more when the cohort is too broad.
  • Follow-up actions are opt-in only.

Warnings

  • Do not guess filter fields or operators.
  • Do not treat lead payments or invoices as if they were the lead record itself.
  • leads_exists_by_email answers existence, not ownership or enrichment.
  • Do not present suspectedFraud as a verdict on the person: it is a heuristic signal, and a missing document alone does not trigger it.

Anti-patterns

  • Asking the user for workspace id.
  • Using raw origins/UTM helpers as a substitute for lead search.
  • Returning every field when the user only asked for one operational answer.
  • Calling leads_validate_email on many leads to "clean the base" (paid, rate-limited) or just to read a status that leads_get already returns.

Clickmax skill revision: a13590fbff81

Add the canonical catalog link to the repository README so users can inspect current installs and available audits. The publishing guide covers the complete discovery path.

<a href="https://skillzs.dev/skills/docs.clickmax.io/clickmax-leads">View clickmax-leads on skillZs</a>