Open beta · one region · no account

Context storage for coding agents

Index a repository once. Your agent pulls back the parts it needs — the matching chunks, their embeddings, and the state it left behind — as a single packed bundle.

Get a token

No sign-up, no email. The token is the identity — it is generated in your browser and shown once.

Index

Stream chunks and their embeddings into a collection over one long-lived connection.

Search

Nearest-neighbour search returns a bundle handle instead of a wall of JSON.

Resume

Fetch the bundle, decode it, and hand it to the agent as working context.

Three calls end to end

export RP_TOKEN=rp_...

# 1. create a collection
curl -X POST https://rebornpad.dev/api/v1/collections \
  -H "Authorization: Bearer $RP_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"name":"monorepo/api","dims":1024}'

# 2. search it
curl -X POST https://rebornpad.dev/api/v1/search \
  -H "Authorization: Bearer $RP_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"collection":"col_9fK2mQ","vector":[...],"k":24}'

# 3. pull the bundle
curl https://rebornpad.dev/api/v1/query/b_4Kd82nQmZs \
  -H "Authorization: Bearer $RP_TOKEN"

The bundle comes back deflated and base64-encoded, so it drops straight into a context loader without parsing. Full details in the API reference.

What the beta gives you

Ingest volume	25 GB per day
Collections	20 per token
Chunks	2,000,000 per collection
Transport	HTTP/3 where available, HTTP/2 otherwise
Region	eu-north
Price	Free during beta

This is a beta, run by two people. There is no SLA and no support rotation. Collections can be pruned with notice if we need the capacity back. Do not put anything here that you cannot re-index from source.

What we store

Only what you send: chunk text, vectors, and the metadata you attach. No accounts means no names, no emails, no addresses. Request bodies are not logged — if something breaks, the X-Request-Id header is all we can trace it by. Details in the privacy note.