LinkedIn, as tools your agent already knows how to call.
One config block.
{
"mcpServers": {
"linkfetch": {
"command": "npx",
"args": ["-y", "linkfetch-mcp"],
"env": { "LINKFETCH_API_KEY": "sk_live_…" }
}
}
}What the agent can call.
| tool | what it does | credits | needs session |
|---|---|---|---|
| linkfetch_search_jobs | Keyword and filter search over the jobs index: location or geo ID, company, level, salary floor, posting window. | 1 | no |
| linkfetch_get_job | Full detail for one posting by LinkedIn job ID — description, parsed salary, applicant count, lifecycle timestamps. | 1 | no |
| linkfetch_get_job_by_url | Same as above from any LinkedIn job URL or URN. | 1 | no |
| linkfetch_get_job_applicants | The timestamped applicant-count series for one posting — how fast interest is moving. | 1 | no |
| linkfetch_jobs_database_info | Coverage, freshness and schema of the jobs index so the agent can reason about what it is looking at. | 0 | no |
| linkfetch_search_locations | Resolve a place name to a canonical LinkedIn geo ID for precise filtering. | 1 | no |
| linkfetch_get_profile | Typed person record by URL or slug: headline, roles, education, skills. | 5 | extension |
| linkfetch_get_company | Company record by slug: industry, headcount band, locations, followers. | 3 | extension |
| linkfetch_get_company_employees | Paginated employee list for a company. | 5 | extension |
Things to say to it.
- “Find every staff-level backend role in Berlin posted this week. Group by company and rank by applicant velocity.”
- “Which fintech companies in the jobs index started hiring machine-learning engineers in the last 30 days? Give me the five with the most open roles.”
- “Look up the company behind this job posting URL and tell me whether it is growing, based on how many roles it has opened this month.”
More in the recipes, each with the tool chain and credit cost.
Questions before you install.
What is the LinkFetch MCP server?
A Model Context Protocol server, published as the linkfetch-mcp npm package, that exposes LinkFetch's LinkedIn API as typed tools. Any MCP client — Claude Desktop, Claude Code, Cursor, Windsurf, Zed, Continue — can call it natively without HTTP plumbing.
Does it need anything besides an API key?
The jobs and location tools read from LinkFetch's own index and need only the key. Profile and company tools resolve through the user's own LinkedIn session via the LinkFetch Chrome extension; on a cache miss they return extension_required so the agent can ask the user to open the page.
What does it cost?
The same as the REST API: flat credits per call, empty results free, and a $5 free credit on sign-up. Starter is $29/month and Pro is $99/month. If your agent needs the whole archive rather than lookups, the jobs dataset is $199 one-time.
Is it hosted, or does it run locally?
It runs locally over stdio and talks to api.linkfetch.io with your bearer key. There is no hosted endpoint to point a client at today.
Is the data legal to use in an agent?
Jobs come from LinkedIn's public guest endpoints — listings any logged-out visitor can see. Profiles and companies are fetched in the user's own session, so the user is always the principal. Every result carries a provenance stamp the agent can cite.