/mcp

Your career context,served over MCP

ProfileClaw runs an MCP server that exposes your consent-scoped career context layer to any MCP client — Claude Code, Cursor, ChatGPT, and more. One tool, four task views, never a raw answer.

What your agent can read

One tool — get_context — returns exactly the slice of your profile the task needs.

Career context

RIASEC, Big Five, values, and skills scores on the 1–5 scale, with the top signals for the task at hand.

Shaped per task

task=career, projects, learning, or communication — each read returns only what that decision needs.

Sourced, never raw

Every score traces back to the dimensions behind it. Raw assessment answers are never exposed.

Connect in one line

Point any MCP client at the ProfileClaw server and authorize with a ProfileClaw key.

claude mcp add --transport http profileclaw \

https://profileclaw.com/api/mcp \

--header "Authorization: Bearer pc_YOUR_KEY"

Endpoint · /api/mcp·Tool · get_context·Scopes · profile / assessments / evidence

You stay in control

Scoped grants

Authorize exactly the layers an agent may read — profile, assessments, evidence — for 7, 30, or 90 days.

Revoke anytime

One click ends a grant, and every read is logged to your audit trail.

Privacy by design

Raw answers stay yours. Agents read derived, consent-scoped context only.

Give your agent real context in one line.

01

Create a key

Issued in your dashboard, scoped per use

02

Add the server

One command in any MCP client

03

Agents start reading

Scoped, sourced, audit-logged

MCP · /api/mcp·Tool · get_context·Scope-level consent