Intakra
ให้คะแนนสัญญาณการซื้อที่ได้รับ และจัดทำสรุปเหตุผลที่ควรซื้อในตอนนี้แบบรีวิวเท่านั้น
เซิร์ฟเวอร์ MCP แบบโฮสต์
npx add-mcp 'https://intakra.com/mcp'ติดตั้งลง Claude Code, Codex, Cursor และอื่นๆ
เอกสาร
Ask your assistant who is worth a look.
Connect to https://intakra.com/api/mcp with OAuth. Sign in to Intakra, choose your workspace and review the permissions. Then ask about the companies on your list:
- “Here is my account list. Who should I target?”
- “Anything happening on my account list this week?”
- “Are any of these companies showing buying signals?”
You get saved findings with dates, source links and a rough time frame. Accounts awaiting checks, missing from your list or without saved signals remain separate. A public event can suggest a reason to investigate; it does not confirm need, budget or purchase intent. Read calls do not run new research.
Reading accounts, adding domains and replacing what you sell use separate permissions. New accounts wait for the normal checking pass under your existing plan. This connection does not expose outreach, contact or CRM tools.
Public MCP
Check the proof on a company from your AI assistant.
Connect an OpenAI or Claude MCP client to https://intakra.com/mcp. These free tools only look at what you paste in, and every result is a draft for you to review. Intakra customers can also connect their account and ask about the companies on their list; that connector lives at the signed-in /api/mcp endpoint.
Available tools
score_buying_signal_evidence
You paste in what you know about a company and what you sell. It returns a simple scorecard: how well they fit, how fresh the proof is, how strong it is, and a rough priority. For you to review, nothing more.
prepare_why_now_review
Turns the same facts into a short note on why now might be a good time to reach out. Every fact stays marked as something you supplied and we have not checked. It never writes outreach.
What these tools do not do
- No Intakra login or API key is required.
- No workspace, CRM, contact, account, or customer data is read.
- No source is opened and no model or paid provider is called.
- No input or result is persisted by the MCP implementation.
- No outreach is drafted or sent, and no CRM action is available.
- Every result is a draft for you to review, not a prediction that anyone will buy.
Keep private workspace data in the signed-in /api/mcp connector. Do not paste personal, confidential, or login data into the public endpoint.