API Reference¶

Every RAG pipeline in hemlock-lab exposes the same 4 HTTP endpoints. This section documents the request/response format for each endpoint.

Base URLs¶

Framework	Base URL
LangChain	`http://localhost:8100`
LlamaIndex	`http://localhost:8101`
Unstructured	`http://localhost:8102`
Haystack	`http://localhost:8103`
ColPALI	`http://localhost:8104`
ChromaDB	`http://localhost:8000`
Ollama	`http://localhost:11434`

Pipeline Endpoints¶

All 5 RAG pipelines expose these endpoints:

Endpoint	Method	Page
`/health`	GET	Health
`/extract`	POST	Extract
`/ingest`	POST	Ingest
`/query`	POST	Query

Quick Reference¶

Health Check (Any Pipeline)¶

curl http://localhost:8100/health

Extract Text¶

curl -X POST http://localhost:8100/extract \
  -F "file=@document.html"

Ingest Document¶

curl -X POST http://localhost:8100/ingest \
  -F "file=@document.html" \
  -F "collection=my-collection"

Query RAG Chain¶

curl -X POST http://localhost:8100/query \
  -H "Content-Type: application/json" \
  -d '{"query": "What is the refund policy?", "collection": "my-collection"}'

ChromaDB Native API¶

For direct ChromaDB access (bypassing the pipeline wrappers):

Endpoint	Method	Purpose
`/api/v1/heartbeat`	GET	Health check
`/api/v1/collections`	GET	List collections
`/api/v1/collections`	POST	Create collection
`/api/v1/collections/{name}`	DELETE	Delete collection
`/api/v1/collections/{id}/add`	POST	Add documents
`/api/v1/collections/{id}/query`	POST	Query by embedding

Examples¶

# List collections
curl http://localhost:8000/api/v1/collections | jq '.[].name'

# Count documents in a collection
curl http://localhost:8000/api/v1/collections/noise-corpus/count

Ollama Native API¶

Endpoint	Method	Purpose
`/api/tags`	GET	List models
`/api/embed`	POST	Generate embeddings
`/api/generate`	POST	Generate text

Examples¶

# List models
curl http://localhost:11434/api/tags | jq '.models[].name'

# Generate embedding
curl http://localhost:11434/api/embed \
  -d '{"model": "nomic-embed-text", "input": "test query"}'

# Generate text
curl http://localhost:11434/api/generate \
  -d '{"model": "smollm2:135m", "prompt": "Hello", "stream": false}'

Error Responses¶

All endpoints return errors in this format:

{
  "error": "description of what went wrong",
  "detail": "optional stack trace or additional info"
}

Common HTTP status codes:

Code	Meaning
200	Success
400	Bad request (missing file, invalid JSON)
415	Unsupported file format
500	Internal server error (framework exception)
503	Service unavailable (ChromaDB or Ollama down)

Next Steps¶

Extract — Full /extract endpoint documentation
Pipelines — How each framework implements these endpoints