X (Twitter) profile qualifier and enricher

You are helping me build an X lead qualifier + enricher. The agent takes a single X username as input, fetches the profile, applies hard disqualify rules, scores against my ICP rubric, and (only for high-fit leads) resolves a LinkedIn URL and finds + verifies a contact email. It returns a structured JSON result.

REFERENCE DOCS (read these before writing code)
- Full agntdata API documentation: https://agnt.mintlify.app/apis/overview
- Endpoints we'll use (all behind one agntdata key):
  - `twitter154 / User_Details` — fetch X profile (bio, location, link, followers)
  - `twitter154 / User_s_Tweets` — fetch recent posts for content scoring
  - `linkedin-data-api / Get_Profile_Data_By_URL` — fetch LinkedIn profile after we resolve the URL
  - `agntdata-agnt / people_email_finder` — find an email from {linkedin_url} or {full_name + domain}
  - `agntdata-agnt / people_email_verifier` — verify deliverability before we surface the email
- Plus the Exa connector for LinkedIn URL discovery and website summarization.

ABOUT MY PRODUCT
- Product: <YOUR PRODUCT — one sentence>
- Top ICPs (in priority order, with a 1-sentence "what they look like on X"):
  1. <ICP 1>: <signals>
  2. <ICP 2>: <signals>
  3. <ICP 3 — optional>
- Tool stack mentions worth boosting (competitive or complementary tools — bio + recent posts):
  - <tool 1, tool 2, tool 3, ...>
- Hard disqualify rules (in priority order; first hit wins):
  1. <rule 1 — e.g. "no link in profile">
  2. <rule 2 — e.g. "no bio">
  3. <rule 3 — e.g. geographic / role-based filters>

WHAT TO BUILD
An agent (claude-haiku-4-5) on agntdata. It accepts an X username as input — from chat, a webhook payload, or another agent's tool call — and runs the pipeline once per invocation.

Step 1 — Parse input. Pull the X username from the user message (strip leading @, trim). If missing or malformed, respond "expected an X username" and stop.

Step 2 — Fetch X profile via `User_Details`. Hold the full response in scratchpad memory.

Step 3 — Hard disqualify gates (apply IN ORDER, stop on first hit). Use my rules above. If any fires: emit a structured "disqualified" result and stop. Do not run any further calls.

Step 4 — Run `fetch_x_recent_posts` skill ({ x_username }). It returns the recent tweets payload.

Step 5 — Run `fetch_website_summary` skill ({ website_url }) using the URL from the X profile. It returns a one-paragraph summary via Exa.

Step 6 — Run `score_lead` skill ({ x_profile, recent_posts, website_summary }). It calls Claude Sonnet with my rubric and returns `lead_score` (0–100), `score_breakdown` (jsonb), `icp_match`, `reasoning`, and `qualification_notes` (only when score > 50).

Step 7 — If `lead_score <= 50`: emit the structured "scored, not enriched" result and stop.

Step 8 — Enrichment (only when `lead_score > 50`):
  8A. Parse `first_name` / `last_name` from the X display name. Extract `company_url` from the X profile link and `company_description` from the website summary.
  8B. Search Exa for LinkedIn: `"<full name>" site:linkedin.com/in`. If a confidently-matching `linkedin.com/in/...` URL is returned, capture it.
  8C. If we have a LinkedIn URL: call `Get_Profile_Data_By_URL` and capture the response.
  8D. Email finder via `people_email_finder`: prefer `{ linkedin_url }`, fall back to `{ full_name, domain }`. `max_cost_cents: 50`. If neither viable: skip to the result with `email_enrichment_status = 'not_found'`.
  8E. **Mandatory verify** via `people_email_verifier`. Only surface the email if status is `deliverable` (`email_verified = true`) or `risky` (`email_verified = false`). Discard `undeliverable`.

Step 9 — Emit the structured result. Always include:
```json
{
  "x_username": "...",
  "disqualified": false,
  "disqualify_reason": null,
  "lead_score": 72,
  "icp_match": "<matched ICP>",
  "score_breakdown": { ... },
  "qualification_notes": "...",
  "first_name": "...",
  "last_name": "...",
  "company_url": "...",
  "company_description": "...",
  "linkedin_profile_url": "...",
  "linkedin_profile_details": { ... },
  "website_summary": "...",
  "email": "...",
  "email_verified": true,
  "email_enrichment_status": "found"
}
```

GUARDRAILS
- Never call X or LinkedIn directly — always go through the agntdata endpoints.
- Don't compute `lead_score` / `score_breakdown` from the agent — only the `score_lead` skill does.
- Verifier is mandatory before surfacing any email. No exceptions.
- Skip the email finder if you only have a first name and no domain — don't waste a call.

RECOMMENDED PIPELINE (optional)
This agent is a function: `x_username → qualified lead`. To turn it into a pipeline, pair it with the `x-engager-scraper` blueprint (which fills a `leads` table with X usernames from engaged threads) and add a thin upstream agent that polls the table and calls this one per row. Or simply pipe the structured result into a webhook to your CRM.

When ready, start by asking me the ABOUT MY PRODUCT questions.