Pre-built workflows for GTM teams using CatchIntent with AI tools like Claude Code, Cursor, and Windsurf.
claude mcp add catchintent --transport http https://engine.catchintent.com/mcpSee setup docs for other clients.
npx skills add catchintent/skillsCatchIntent has one unified surface: agents create leads you triage, research, and act on. Every lead is a person on either LinkedIn or X — the sourcePlatform field on each lead tells you which. These skills cover the discovery side (agents → leads → research). Outreach skills will land once the outreach module ships.
| Skill | What it does | Trigger |
|---|---|---|
| daily-triage | Walk the new-lead queue one at a time — save / archive / skip until clear | "Triage my leads", "what's new", "clear the queue" |
| research-lead | Deep-research a single LinkedIn lead — public sources first, paid enrichment only if approved | "Research X", "what's the angle on this lead" |
| enrich-leads | Bulk paid enrichment via Apify ($0.002/lead, idempotent) for LinkedIn leads | "Enrich these leads", "fill in profile data" |
| tune-agent | Diagnose an underperforming agent and propose specific targeting changes | "Tune my agent", "leads are off" |
| pipeline-review | Pipeline health check — usage, agents, lead funnel, queue; monthly mode adds a lead-quality retro | "How's my pipeline", "weekly review", "monthly retro" |
| craft-offer | Sharpen the structured offer (pain point, mechanism, proof) that powers AI lead scoring | "Fix my offer", "what's my one-liner" |
Daily:
/daily-triage → /research-lead (on the high-priority saved leads) → /enrich-leads (LinkedIn leads missing detail)
When leads feel off:
/daily-triage → /tune-agent → wait 3-5 days → /daily-triage
First-time setup:
/craft-offer → create your first agent → /daily-triage
Weekly / monthly:
/pipeline-review → /tune-agent (on the worst agent)
Skills are playbooks that tell your AI tool how to compose CatchIntent's MCP tools (agent_*, lead_*, engagement_*, product_*, workspace_*, denylist_*) into useful workflows. The tool surface is unified — there are no SL- or LI-specific tool names. Where lead shape matters, tools take a kind discriminator.
All skills follow STANDARDS.md — shared conventions for data-first analysis, confirm-before-write, and cross-skill references.
See STANDARDS.md for skill authoring conventions and the agentskills.io spec for the SKILL.md format.