For recruiters
Search, rank, and evaluate
inside your AI tools
No new portal, no new login. One config line connects Claude, ChatGPT, or Cursor to candidates who put their own career data here and keep it current.
Thirty seconds of setup
Add this to your client's MCP settings, restart it, and ask your first question. The same config works in Claude Desktop, ChatGPT developer mode, and Cursor.
{
"mcpServers": {
"restume": {
"url": "https://mcp.restume.com/mcp"
}
}
}
What a real call returns
This is an actual request and the actual response from the live server, trimmed for width. Every score has a visible breakdown; nothing is a black box.
> rank_candidates_for_job(
job_description="Junior full-stack engineer for a fintech startup.
Must know Python, deploy with Docker. Kubernetes a plus.",
required_skills=["Python", "Docker"],
preferred_skills=["Kubernetes", "FastAPI"])
{
"candidate": {
"username": "hershelthomas",
"open_to_work": true,
"skills_count": 30, "experience_count": 3, "projects_count": 6
},
"fit": {
"total_score": 91.16,
"matched_required_skills": ["python", "docker"],
"missing_required_skills": [],
"matched_preferred_skills": ["kubernetes", "fastapi"],
"experience_years": 2.2,
"breakdown": {
"required_skills_score": 55.0,
"preferred_skills_score": 20.0,
"job_keywords_score": 2.86,
"experience_score": 3.3,
"completeness_score": 10.0
}
}
}
The four calls your agent will actually make
search_candidates
Filter by skills, location, companies, keywords, and minimum experience. Returns scored candidates with a five-part relevance breakdown.
rank_candidates_for_job
Paste a job description, get every candidate ranked with matched and missing skills spelled out per candidate.
explain_candidate_fit
A strong-fit, moderate-fit, or weak-fit verdict with risk flags like required-skill gaps or thin profiles.
answer_profile_question
Ask "does he actually know Kubernetes?" and get the structured evidence behind the answer, not a summary you have to trust.
Eleven more tools cover comparison, shortlisting, skill-gap analysis, and full profile reads. The MCP reference lists them all.
Data candidates chose to give you
Sourcing tools sell you scraped profiles that go stale the day they are indexed. Every profile here was created and is maintained by the candidate, and answers cite the specific roles, projects, and skills they come from.
There is no seat license. Querying candidates over MCP is free, and so is the profile every candidate gets.