llms.txt

# DocForge

> Document → Clean Markdown / Chunks for RAG + agent ingest

**GitHub:** https://github.com/bdeeps/docforge  
**Live demo:** https://docforge.konsole.one

DocForge converts PDF, HTML, DOCX, and text into normalized Markdown plus deterministic chunked JSON with token estimates.

## Start here (agents)

- [Agent guide (AGENTS.md)](/AGENTS.md): self-registration, MCP, REST, chunking
- [Agent manifest (JSON)](/api/v1/agent-manifest): machine-readable discovery
- [Cursor skill](/docs/skill.md): step-by-step agent instructions

## Documentation

- [FAQ (AEO)](/docs/faq.md): 24 Q&A — formats, chunking, registration, MCP, troubleshooting
- [MCP reference](/docs/agents/mcp.md)
- [REST API reference](/docs/agents/rest.md)
- [Chunking strategies](/docs/agents/chunking.md)
- [OpenAPI interactive docs](/api/docs)

## Self-register

```http
POST /api/v1/agents/register
Content-Type: application/json

{ "name": "your-agent-name", "capabilities": ["ingest", "chunk"] }
```

Save `api_key` from the response (shown once).

## Ingest a document

```http
POST /api/v1/ingest
X-DocForge-Key: df_YOUR_KEY
Content-Type: multipart/form-data

file=@document.pdf
strategy=headings
max_tokens=512
```

## MCP

Command: `docforge-mcp`  
Tools: `register_agent`, `ingest_document`, `chunk_markdown`, `convert_document`, `get_agent_manifest`, `agent_whoami`, `health_check`

## Optional

- [Human UI](/): upload, chunk config, preview
- [Well-known manifest](/.well-known/docforge.json)