hrms-linkedin-mentor

name	hrms-linkedin-mentor
description	Import LinkedIn profiles into IntelliForge mentor accounts for internal internship programs. Use when building or debugging LinkedIn-to-mentor workflows, /api/mentors/import-linkedin, or onboarding mentors from social profiles.

LinkedIn → Mentor Profile Workflow

What it does

Turns a LinkedIn profile into a MentorProfile (and optionally a new Admin with MENTOR role) using OpenAI structured extraction.

In-app entry points

Who	Where	Action
Mentor (self)	/dashboard/mentor-profile	Import from LinkedIn → action: apply-self
Org admin	/dashboard/mentors/import	Create mentor from LinkedIn → action: create-mentor

API

POST /api/mentors/import-linkedin (authenticated)

Preview (mentor or admin)

{
  "action": "preview",
  "linkedinUrl": "https://www.linkedin.com/in/jane-doe/",
  "profileText": "optional pasted About + Experience"
}

Returns { draft, source, warning? }.

Apply to signed-in mentor

{
  "action": "apply-self",
  "linkedinUrl": "...",
  "profileText": "...",
  "isPublic": false
}

Upserts MentorProfile for the current admin.

Create mentor + profile (full admin only)

{
  "action": "create-mentor",
  "linkedinUrl": "...",
  "profileText": "...",
  "email": "[e-mail]",
  "password": "********",
  "confirmPassword": "********",
  "sendWelcomeEmail": true,
  "isPublic": false
}

1. assertCanAddMentor(orgId)
2. createOrgAdminDirect with MENTOR role
3. upsertMentorProfileForAdmin with extracted fields

LinkedIn fetch reality

LinkedIn blocks most server-side fetches. Always recommend pasting profile text (About, headline, skills, experience). The UI shows a warning when only URL-only extraction runs.

Key files

• src/lib/ai/linkedin-mentor-import.ts — URL normalize + OpenAI extraction
• src/app/api/mentors/import-linkedin/route.ts — preview / apply / create
• src/components/mentors/linkedin-import-form.tsx — shared UI
• src/lib/validations.ts — linkedInImportPreviewSchema, linkedInImportApplySchema, linkedInImportCreateSchema

Env

• OPENAI_API_KEY — required for extraction

Agent workflow (Cursor)

When the user provides a LinkedIn URL in chat:

1. Ask them to paste About + Experience if the draft looks thin.
2. Call preview mentally or via API locally.
3. For production onboarding, direct them to /dashboard/mentors/import or use the API with admin session.
4. Never scrape LinkedIn aggressively; respect ToS — pasted text is the supported path.

MentorProfile fields mapped

LinkedIn signal	DB field
Name	Admin.name (create flow)
Headline	headline
About + roles	bio
Skills / stack	expertise[]
Career length	yearsExperience
Profile URL	linkedinUrl
GitHub (if visible)	githubUrl

Testing

npx vitest run tests/unit/linkedin-mentor-import.test.ts