agent mode

let your AI do the networking

connect Cursor or Claude Desktop. your agent searches talent, ranks matches, and stays out of your DMs.

demo keycopy paste don't gatekeep

sh_demo_sportshire_arpand9

cursor config

{
  "mcpServers": {
    "sportshire": {
      "command": "npx",
      "args": ["tsx", "apps/mcp-server/src/index.ts"],
      "cwd": "/path/to/sports-jobs-agent",
      "env": {
        "DATABASE_URL": "postgresql://sportshire:sportshire@localhost:5432/sportshire",
        "SPORTSHIRE_API_KEY": "sh_demo_sportshire_arpand9",
        "SPORTSHIRE_SEEKER_PRO": "0",
        "SPORTSHIRE_EMPLOYER_PRO": "0"
      }
    }
  }
}

agent playbook

1
call search_jobs — query "Sports Analytics Manager"
2
call find_best_jobs with seeker_name: "Sarah Chen" — free tier, full scores
2b
set SPORTSHIRE_EMPLOYER_PRO=1 then find_best_candidatesfeatured job_id: d94ce7e3-422e-46ea-a5cd-f45d0887e156
3
call get_candidate_profile on #1 — main character unlocked

test in browser

run the same tools against your local DB without MCP — open live tools playground →

free vs pro tools

Toggle Pro in the web demo at /demo or set env vars in MCP config.

Always free

search_jobs
search_candidates
find_best_jobs
get_job_posting
get_candidate_profile
get_marketplace_stats

Pro (env flag)

apply_to_job — SPORTSHIRE_SEEKER_PRO=1
find_best_candidates — SPORTSHIRE_EMPLOYER_PRO=1
match_candidate_to_job — SPORTSHIRE_EMPLOYER_PRO=1
get_match_results (by job_id) — SPORTSHIRE_EMPLOYER_PRO=1