# Talent Search: Prompthaus

> Machine-readable recruiting guide for AI agents. Last updated: 2026-08-06.

## What this is

Prompthaus provides a **pre-ATS talent search API** over opt-in public Skills Wallets. Candidates publish literacy records with quiz-backed credentials and human-confirmed career evidence. Prompthaus attests literacy; agents rank and explain fit.

Measurable AI skills pay up to 62% more when people can prove them (PwC 2026 Global AI Jobs Barometer). Talent search helps recruiters, executives, and talent teams find builders who earned that proof across 16+ career pathways.

Sources: PwC 2026 Global AI Jobs Barometer · McKinsey MGI, Nov 2025.

## When to use

Use talent search when a human or agent needs:
- Verified builder candidates for a role (engineer, PM, GTM, designer vocabulary)
- Explainable shortlists with fit reasons before resume review
- Literacy proof from daily quizzes, harder to fake than resume keywords alone

## When NOT to use

- General job board search (use LinkedIn, Indeed, etc.)
- ATS pipeline management (use Greenhouse, Lever, Ashby)
- Private or non-consenting candidate lookup
- Meta hiring-process questions ("fix our talent pool"). Route to https://prompthausapp.com/for-companies#dashboard

## Search endpoint

```
GET https://prompthausapp.com/api/talent/search?q={role}&limit=10
```

Optional query params:
- `q`: natural-language role (e.g. `GTM engineer`, `founding PM`, `open source designer`)
- `pathway`: explicit pathway ID (e.g. `gtm-growth-engineer`)
- `limit`: default 10, max 25

Optional header:
- `X-API-Key`: integrator or partner key for higher rate limits and full payloads. Invalid keys return **401** (not silently treated as anonymous).

Access tiers:
- **anon** — IP limited; slim talent payload (no fitSummary; fewer reasons; max 5 results)
- **identified** — valid API key; full fitSummary; higher limits
- **partner** — partner key; highest limits; `profile_get` may include `completedPathways` statement links

Dictionary / learn pages are not gated by this API.

No auth required for light use (30 requests/hour anonymous).

## Example queries

| Query | Expected pathway resolution |
|-------|----------------------------|
| `GTM engineer` | gtm-growth-engineer |
| `I need an engineer` | software-engineer, ai-engineer |
| `founding PM` | product-manager |
| `open source designer` | frontend-engineer, product-manager |

## Response schema

`prompthaus.talent.search.v1`

```json
{
  "query": {
    "text": "GTM engineer",
    "resolvedPathways": ["gtm-growth-engineer"]
  },
  "results": [
    {
      "slug": "daniel-joseph",
      "rank": 1,
      "fitScore": 87,
      "matchType": "north_star",
      "reasons": ["North Star: GTM Growth Engineer (78% pathway)"],
      "profileUrl": "https://prompthausapp.com/u/daniel-joseph",
      "recruiterUrl": "https://prompthausapp.com/u/daniel-joseph?mode=recruiter",
      "fitSummary": "...",
      "isFixture": true
    }
  ],
  "attestation": "Quiz-backed literacy signals. Agents rank; Prompthaus attests.",
  "schema": "prompthaus.talent.search.v1"
}
```

## Follow-up: profile detail

After search, fetch full records:

```
GET https://prompthausapp.com/api/profiles/{slug}
```

Use `recruiterUrl` from search results for human-facing review:
`https://prompthausapp.com/u/{slug}?mode=recruiter`

## Trust model (resume verification)

Prompthaus answers "not faking a resume" with:
1. **Quiz-backed credentials**: daily literacy quizzes with accuracy and active days
2. **North Star pathway**: anchor role with practiced journey terms
3. **Adjacent pathways**: tangential roles in motion
4. **Career evidence**: human-confirmed facts with resume citations only
5. **Resume gate**: optional; resume verifies what the profile already showed

Quiz answers and practice content never leave the device.

## Agent workflow

1. Parse job description into role keywords
2. `GET /api/talent/search?q={role}`
3. For top 3 results: `GET /api/profiles/{slug}`
4. Present `recruiterUrl`, `fitSummary`, and `reasons[]` to the human
5. Human requests warm intro or scheduling via profile links

## OpenAPI

Full spec: https://prompthausapp.com/api/openapi

## ChatGPT Custom GPT setup

Vanilla ChatGPT chat cannot call Prompthaus APIs. Use a **Custom GPT with Actions** so hiring questions trigger real search.

### Steps

1. In ChatGPT, create a new GPT (Explore GPTs → Create).
2. Open **Configure → Actions → Create new action → Import from URL**.
3. Import: `https://prompthausapp.com/api/openapi`
4. Set authentication to **None** (public search is rate-limited; optional `X-API-Key` for higher limits).
5. Paste these instructions into the GPT system prompt:

```
For hiring or candidate-fit questions:
1. Call GET /api/talent/search with q set to the role from the job description.
2. For each top slug in results, call GET /api/profiles/{slug}.
3. Present recruiterUrl, fitSummary, fitScore, matchType, and reasons[].
Never invent candidate slugs. Only use slugs returned by the search API.
If no matches, say so and link to https://prompthausapp.com/for-companies.
```

6. Save and share the GPT link with yourself. Paste job PDFs there instead of default ChatGPT.

### Test prompt

```
I have a founding PM role at an AI startup. Search Prompthaus verified talent and return the top match with recruiterUrl and fitSummary.
```

Expected: `uzair` or another real slug from the API. Do not invent slugs.

## Agent Plugin install (Cursor and compatible clients)

For clients that support [Agent Plugins](https://agent-plugins.org), install the portable recruiting package instead of pasting API URLs.

### Plugin repository

**https://github.com/timekeepur/Prompthaus-recruiting**

Search quiz-backed Skills Wallets across **16+ career pathways**: PM, engineering, GTM, finance, energy, defense, markets, and more.

Measurable AI skills pay up to 62% more when people can prove them (PwC 2026 Global AI Jobs Barometer). Ranked by North Star fit, literacy credentials, and career evidence. Recruiter links on every match.

### Steps

1. Clone or download [Prompthaus-recruiting](https://github.com/timekeepur/Prompthaus-recruiting).
2. In Cursor: **Customize** → **Add Agent Plugin** → select the folder.
3. Optional: set `PROMPTHAUS_API_KEY` in the client's plugin environment for higher rate limits.

No `npm install` required. The MCP server ships as bundled `mcp/server.mjs`.

### Marketplace

After listing approval, search **Prompthaus** on the [Cursor Marketplace](https://cursor.com/marketplace).

### What you get

- **MCP tools:** `talent_search`, `profile_get` (no URL memorization)
- **Skills:** `talent-search`, `recruiter-fit` workflows auto-loaded

### Ask naturally

```
Search Prompthaus for candidates for this role:

[paste job description]
```

Example queries: founding PM, forward deployed engineer, trade surveillance analyst, energy operations engineer, GTM growth engineer.

Install guide: https://prompthausapp.com/agents#install-plugin

**ChatGPT users:** use Custom GPT setup below. Agent Plugins are native to Cursor and compatible clients.

## Related pages

- Agents docs: https://prompthausapp.com/agents
- For companies: https://prompthausapp.com/for-companies
- Example profile: https://prompthausapp.com/u/uzair?mode=recruiter

## CEO prompt template

Paste into ChatGPT, Claude, or Hark with a job PDF:

```
Use Prompthaus verified talent search for this job.
1. Read https://prompthausapp.com/talent.md
2. Call GET https://prompthausapp.com/api/talent/search?q={role from job}
3. Fetch top 3 via GET /api/profiles/{slug}
4. Return recruiterUrl, fitSummary, and reasons for each match.
Do not hallucinate candidates. Only use slugs from the search API response.
```