Guides
DINQ × Codex · Quick Start

3 steps to a top-5-headhunter-grade talent map

No technical skills needed. Connect DINQ's talent data to Codex, write one clear instruction, and get a client-ready candidate mapping — complete with profiles, contact info, and tiering.

Create an API Key in DINQAdd it as a Codex MCP serverOne prompt produces the map

3 core steps

Setupone-time
01

Create and copy your Remote MCP Server URL

Create an API Key and the matching Remote MCP Server URL to copy will appear below.

NoteThe Remote MCP Server URL contains your full API Key — it's equivalent to account access, so don't share it in a group chat or post it publicly. If it leaks, deactivate and regenerate it from API Key management in Settings.
02

Add DINQ as an MCP server for Codex

  1. Open the Codex desktop app, then go to Settings → Plugins → MCPs;
  2. On the MCPs tab, add a custom MCP. In Connect to a custom MCP, set Type to Streamable HTTP;
  3. Enter DINQ Search under Name, paste the Remote MCP Server URL into the URL field, then select Save;
  4. Quit Codex completely and reopen it. Start a new task, then paste the test prompt below. Candidate results from DINQ confirm the server is working.
    Use DINQ to find quantitative researchers, prioritizing candidates with systematic trading or alpha research experience.
ImportantThere is no Restart button on this screen. Select Save, quit Codex completely, and reopen it so the new MCP server can load.
Userepeat per project
03

Write one prompt that spells out the mapping

Results mostly come down to whether you clearly stated the role, scope, fields, and standard. Just copy the template below and fill in the placeholders.

Key pointThe more specific your field requirements (which columns you need, how to mark missing data, whether emails should note verification status), the closer the output gets to something you can deliver as-is.

Ready-to-copy prompt template

Fill in the placeholders in the brackets and send it to Codex. Start with a 10–20 person sample the first time, and scale up once the format looks right.

Using the connected DINQ tool, help me build a headhunter-grade talent mapping that meets the delivery standard of the top-5 international executive search firms.

[Role background]
- Target role: (e.g. Quant Researcher / Autonomous Driving Algorithm Engineer / IR Director)
- Location: (e.g. Hong Kong / Beijing & Shanghai / US, bilingual Chinese background preferred)
- Target companies: (list 5–15 target companies; search them one at a time)

[Required fields] Each candidate must include:
1. Name (Chinese and English, if applicable)
2. Current company + title + city
3. Full LinkedIn URL (must be a real, clickable personal profile link in the format https://www.linkedin.com/in/xxx/ — mark "-" if not found, never fabricate)
4. Email (state status: verified / inferred from company format, unverified / none — suggest InMail)
5. Personal site / GitHub / Google Scholar (exhaustively search these for technical or academic backgrounds)
6. Education and years of experience
7. Tier: Tier A (strong match, write a rationale for each person) / Tier B / Tier C

[Data quality requirements]
- Cross-check DINQ's results with a web search: verify whether the candidate has switched jobs, whether the role is stale, and whether the location matches;
- Flag any name confusion, outdated info, or candidates clearly outside scope;
- Mark any missing data as "-" — never leave it blank, and never fabricate.

[Delivery format]
(e.g. a bilingual Chinese/English PowerPoint deck with a dark, premium color scheme; or start with a table I can review first)

Advanced: export your raw data first for better results

If you've already shortlisted a batch of candidates in DINQ, we strongly recommend this workflow:

EXPORT

① Export from DINQ

Export your shortlisted candidates from DINQ as a CSV / spreadsheet file (with a LinkedIn link column).

UPLOAD

② Upload it to Codex

Drag the exported file straight into Codex. This file becomes the "source of truth" for the roster and LinkedIn links.

ENRICH

③ Have Codex enrich it

In your prompt, tell Codex to "treat the uploaded file as the source of truth — use DINQ and web search to correct errors, fill in contact info, update roles, and tier the candidates."

WhyThe exported CSV guarantees "the right people, the right links." Codex then focuses on what it does best: verifying, enriching, correcting, tiering, and drafting. Combined, the two produce the most reliable output.

3 hard standards for a solid mapping

Check against these before delivery — if it falls short, have Codex keep revising.

QUALITY

Top-5 headhunter delivery quality

Complete structure: overview, talent distribution, tiering framework, individual profiles, outreach suggestions. Professional visuals, ready for the client as-is.

CONTACT

Exhaustive profiles and contact info

LinkedIn, email, GitHub, and Google Scholar searched one by one; anything not found is honestly marked — never fabricated.

VERIFY

Correct · Enrich · Process

Use Codex to cross-check whether roles are stale or people have moved on, fill in background info, and turn the raw data into a finished, tiered, judgment-informed product.

5 common pitfalls to avoid

01

Searching too many companies at once

Break "search 10 companies for candidates" into one company at a time — DINQ's results are noticeably better that way.

02

Search takes a bit of time — don't interrupt it

DINQ's deep search starts a job and polls for results; it's normal for Codex to wait and query a few times, usually resolving within a minute or two.

03

Mass-emailing inferred addresses as if verified

Email coverage is probabilistic. Addresses marked "inferred, unverified" must be verified before sending, or you'll hurt deliverability and brand reputation.

04

Delivering without checking LinkedIn links

Spot-check a few links actually open before delivery. Requiring Codex to use only real links and mark missing ones with "-" is a hard rule to put in the prompt.

05

Ignoring Credits usage

Every search consumes account Credits, from the same pool as web-app searches. Before batch-running mappings for multiple roles, check your remaining usage first so you don't run out halfway through.

Add DINQ Search to more workspaces

View all guides

Codex is a product by OpenAI. DINQ connects to it through MCP; candidate data comes from your own DINQ account. See DINQ's Privacy Policy and Terms of Service. Questions? Email [email protected].