People Research (Exa)

Find people via Exa's LinkedIn-weighted index — by company + role, role + location, or a named individual. Calls the Exa REST API through a local script; no MCP server needed, only an EXA_API_KEY.

Setup (once)

bash
export EXA_API_KEY=your-key            # or EXA_API_KEY=... in a .env at the repo root

Get a key at https://dashboard.exa.ai/api-keys. Shared details: exa-native-base.

Run it

bash
python scripts/people_research.py "<role + company/location>" [-n N] [--json]

Examples:

bash
python scripts/people_research.py "engineer at OpenAI" -n 10
python scripts/people_research.py "Head of Growth B2B SaaS startup San Francisco" -n 12
python scripts/people_research.py "Jane Smith Anthropic machine learning" -n 5

Defaults to category=linkedin profile. Be specific — vague queries like "researcher founder CEO startup" match many irrelevant profiles. Always include a concrete company, timeframe, role, or location.

Token isolation (for team maps / lists)

For comprehensive coverage, dispatch subagents in parallel and merge. Search by department + seniority in parallel rather than one broad query:

bash
python scripts/people_research.py "category:people engineering at Acme"        -n 10
python scripts/people_research.py "category:people product design at Acme"     -n 10
python scripts/people_research.py "category:people sales marketing at Acme"    -n 10

Supplement LinkedIn with non-LinkedIn sources (use exa-native-base's generic exa.py):

bash
python ../exa-native-base/scripts/exa.py search "Acme team page employees about us" -n 5
python ../exa-native-base/scripts/exa.py search "joined Acme recently hired new role announcement" -n 5

Deduplicate by LinkedIn URL (canonical), or by name + current company as a fallback.

Override categories with -c

linkedin profile (default) · personal site (their own writing) · news (press mentions). For outbound prospect lists that start from companies, use exa-lead-generation.

After you get results

• Results are similarity, not validation — confirm role/company from the snippet before reporting.
• Deep-read a profile or bio page: python ../exa-native-base/scripts/exa.py contents <url> --text.
• Deliver a table: name · current role · company · location · LinkedIn URL.