lead-qualifier

A ProAgentStore agent that receives inbound leads via webhook, scores them with Workers AI (hot / warm / cold), stores them in a Durable Object, and fires outbound notifications for hot leads.

Endpoints

Method	Path	Description
POST	/webhook/ingest	Receive a new lead (form submission)
GET	/leads	List all leads — supports ?status= and ?score= filters
GET	/leads/:id	Get a single lead with AI notes
PUT	/leads/:id/status	Manually update lead status
GET	/stats	Aggregate counts and conversion rate

Webhook ingest payload

{
  "name": "Jane Smith",
  "email": "jane@acme.com",
  "company": "Acme Corp",
  "message": "We are evaluating CRM tools for our enterprise team — budget approved.",
  "phone": "+1-555-0100",
  "role": "VP of Sales"
}

name and email are required. All other fields are optional and are stored as-is. Extra fields (phone, role, budget, etc.) are passed to the AI for scoring context.

The response includes the assigned score:

{
  "id": "uuid",
  "score": "hot",
  "scoreValue": 88,
  "status": "new"
}

Lead statuses

new → contacted → qualified → converted or new → disqualified

Update via PUT /leads/:id/status with body { "status": "contacted" }.

Secrets

Set these via wrangler secret put and mirror in Doppler (pags project):

Secret	Required	Description
WEBHOOK_SECRET	No	If set, callers must send X-Webhook-Secret: <value>
NOTIFY_WEBHOOK	No	URL to POST hot lead events to (Slack, n8n, Zapier)

Development

pnpm install
pnpm dev

Test the webhook locally:

curl -X POST http://localhost:8787/webhook/ingest \
  -H "Content-Type: application/json" \
  -d '{"name":"Jane Smith","email":"jane@acme.com","company":"Acme Corp","message":"Ready to buy, budget approved, need enterprise plan."}'

Deploy

pnpm deploy
# or push to main — GitHub Actions auto-deploys

After first deploy, set secrets:

wrangler secret put WEBHOOK_SECRET
wrangler secret put NOTIFY_WEBHOOK

Hot lead notification payload

When a lead scores hot, the agent POSTs to NOTIFY_WEBHOOK:

{
  "event": "hot_lead",
  "lead": {
    "id": "uuid",
    "name": "Jane Smith",
    "email": "jane@acme.com",
    "company": "Acme Corp",
    "message": "...",
    "scoreValue": 88,
    "notes": "Enterprise company with budget approved. Decision maker (VP) ready to buy immediately.",
    "createdAt": "2026-06-06T12:00:00.000Z"
  }
}

Scoring model

Uses @cf/meta/llama-3.3-70b-instruct-fp8-fast via Workers AI. The agent asks the model to output a score (0–100) and classification (hot/warm/cold) as JSON, with 2–3 sentence reasoning. If AI is unavailable, falls back to keyword heuristics.

Thresholds: 70+ = hot, 40–69 = warm, 0–39 = cold.