| name | deepread-agent-setup |
| title | DeepRead OCR — Agent Self-Signup |
| description | Zero-friction agent signup for production OCR. Your AI agent gets its own API key in seconds via device authorization — no dashboard, no copy/paste. 97%+ accuracy with multi-model consensus and human-in-the-loop review. 2,000 free pages/month. |
| disable-model-invocation | true |
| metadata | {"openclaw":{"homepage":"https://www.deepread.tech"}} |
DeepRead OCR — Zero-Friction Agent Setup
DeepRead is an AI-native OCR platform that turns documents into high-accuracy data in minutes. Your agent signs up and gets its own API key automatically — no dashboard, no copy/paste.
Using a multi-pass pipeline (PDF → Convert → Rotate Correction → OCR → Multi-Model Validation → Extract), DeepRead achieves 97%+ accuracy and flags only uncertain fields for Human-in-the-Loop review — reducing manual work from 100% to 5-10%. Zero prompt engineering required.
Core Features
- Agent Self-Signup — your agent gets an API key via device authorization (RFC 8628), you just click Approve
- Text Extraction — convert PDFs and images to clean markdown
- Structured Data — extract JSON fields with confidence scores
- HIL Interface — uncertain fields are flagged (
hil_flag) so only exceptions need manual review
- Multi-Model Consensus — cross-validation between models for reliability
- Blueprints — optimized reusable schemas with 20-30% accuracy improvement
- Free Tier — 2,000 pages/month (no credit card required)
Agent Self-Signup
Your agent gets its own API key automatically. No dashboard, no copy/paste. You just click "Approve" once in the browser. This uses secure device authorization (RFC 8628) — the same pattern used by GitHub CLI, Slack, and VS Code.
Step 1 — Agent requests a device code (no auth needed):
curl -X POST https://api.deepread.tech/v1/agent/device/code \
-H "Content-Type: application/json" \
-d '{"agent_name": "My Agent"}'
Response:
{
"device_code": "a7f3c9d2e1b8...",
"user_code": "HXKP-3MNV",
"verification_uri": "https://www.deepread.tech/activate",
"verification_uri_complete": "https://www.deepread.tech/activate?code=HXKP-3MNV",
"expires_in": 900,
"interval": 5
}
Step 2 — Agent opens verification_uri_complete in the browser. You log in (or sign up — free, no credit card), see the agent name, and click Approve.
Step 3 — Agent polls for the API key every 5 seconds:
curl -X POST https://api.deepread.tech/v1/agent/device/token \
-H "Content-Type: application/json" \
-d '{"device_code": "a7f3c9d2e1b8..."}'
| error | api_key | Meaning |
|---|
"authorization_pending" | null | User hasn't approved yet — keep polling |
null | "sk_live_..." | User approved — save the key |
"access_denied" | null | User denied — stop polling |
"expired_token" | null | Code expired (15 min) — start over |
Step 4 — Agent saves the key. Done — ready to process documents.
Manual Setup (Fallback)
If you prefer to skip the agent flow, get a key from the dashboard and set it manually:
export DEEPREAD_API_KEY="sk_live_your_key_here"
Supported Platforms
DeepRead skills work across all major AI coding agents.
After installing, use /setup (agent signup + first document) and /api (full API reference).
| Platform | Install | Skills |
|---|
| Claude Code | npx skills add deepread-tech/skills | /setup, /api |
| Cursor | Copy .cursor/rules/ from skills repo | Auto-invoked rules |
| Windsurf | Copy .windsurf/rules/ from skills repo | Auto-invoked rules |
| ClawHub | clawhub add DeepRead001/deepread-agent-setup | /setup, /api |
| MCP agents | Use device flow endpoints directly | Full API access |
Quick Example
Once you have an API key (via agent self-signup or manual), process any document:
curl -X POST https://api.deepread.tech/v1/process \
-H "X-API-Key: $DEEPREAD_API_KEY" \
-F "file=@invoice.pdf" \
-F 'schema={
"type": "object",
"properties": {
"vendor": {"type": "string", "description": "Vendor name"},
"total": {"type": "number", "description": "Total amount"},
"date": {"type": "string", "description": "Invoice date"}
}
}'
Processing is async (takes 2-5 minutes). You get a job ID immediately:
{
"id": "550e8400-e29b-41d4-a716-446655440000",
"status": "queued"
}
Then get results via webhook (recommended) or polling GET /v1/jobs/{id}. Final response:
{
"status": "completed",
"result": {
"data": {
"vendor": {"value": "Acme Corp", "hil_flag": false},
"total": {"value": 1250.00, "hil_flag": false},
"date": {"value": "2024-10-??", "hil_flag": true, "reason": "Partially obscured"}
}
},
"preview_url": "https://preview.deepread.tech/abc1234"
hil_flag: false — confident extraction, auto-process
hil_flag: true — uncertain, route to human review (typically only 5-10% of fields)
API Overview
| Endpoint | What it does |
|---|
POST /v1/agent/device/code | Start agent self-signup |
POST /v1/agent/device/token | Poll for API key |
POST /v1/process | Process a document (PDF, JPG, PNG) |
GET /v1/jobs/{id} | Check job status |
GET /v1/blueprints | List optimized schemas |
POST /v1/optimize | Create a new blueprint |
GET /v1/preview/{token} | Public preview link (no auth) |
Base URL: https://api.deepread.tech
Auth: X-API-Key: sk_live_... header
For complete API documentation with all examples, webhooks, blueprints, schema templates, and best practices, see the full API skill: deepread-ocr
Pricing
| Tier | Pages/month | Rate limit | Price |
|---|
| Free | 2,000 | 10 req/min | $0 (no credit card) |
| Pro | 50,000 | 100 req/min | $99/mo |
| Scale | Custom | Custom | Contact us |
Links
Ready to start? Run /setup and your agent handles the rest.