Lead enrichment pipeline

# Lead enrichment pipeline

You have the "lead-scorer" MCP server connected (Lead Scorer CRM — endpoint https://mcp.lead-scorer.com/mcp, Bearer key created in Lead Scorer → Settings). Use its tools for every read and write. Never invent data: if a tool result is empty, say so.

## Goal
Take list <LIST_ID> from raw to campaign-ready: enriched profiles, found emails, and an AI summary per lead.

## Steps
1. **Inventory.** `get_leads_from_list` — count who has a LinkedIn URL, an email, an enriched profile.
2. **Profile + company enrichment.** Call `enrich_leads` lead by lead (small batches; a big batch that half-fails is harder to retry). Enrichment REQUIRES the lead to have a linkedin_url — skip and report those without one. Use `enrich_company` for companies with a LinkedIn but no data.
3. **Contact finding.** `find_lead_contact_info` for leads without an email. Results can be ASYNC: re-read the list later via `get_leads_from_list` instead of assuming failure.
4. **AI layer.** `get_leads_pending_ai_enrichment`, then for each: write a 3-4 line summary and 2-3 actionable insights (angle to open with, risk, timing) from the enriched data, and store them with `submit_lead_ai_enrichment`.
5. **Reconcile.** Enrichment data is the source of truth: if it contradicts what web research said (role changed, company pivoted), update your notes and say so.
6. **Report.** Enriched / emails found / AI-summarized / skipped (and why).

## Hard rules
- One lead at a time on enrichment writes; never fire-and-forget a 500-lead batch.
- No outreach here. This skill ends when the list is ready.