Calypso Buckets

Source memoryfor groundedAI agents.

Create named Buckets for PDFs, docs, websites, screenshots, Q&A, and customer files. Calypso keeps every source scoped, indexed, and ready for agents to search and cite.

Create your first bucket Read API docs

Batch uploads Readiness states Scoped memory

Bucket active

bucket://support-handbook

Ready

Indexed members12 / 12

files

websites

Q&A

Attached agent

calypso-rag-agent:support

Memory scoped to this bucket.

Buckets create a source boundary

Every upload lands in a named source space before it becomes searchable by an agent.

Uploads arrive with intent

Files, websites, batches, and MCP uploads target a bucket id, slug, or name before indexing starts.

Readiness is explicit

Accepted, queued, indexed, and bucket-active states make retrieval readiness visible before tests run.

Agents search curated memory

Attach selected Buckets to hosted agents and profiles so retrieval stays focused and repeatable.

Source boundaries

Give every agent the right memory.

Buckets keep support docs, pricing rules, legal policies, onboarding guides, screenshots, and customer files separated by purpose — so agents search the source set they are supposed to trust.

support-handbook

Support

Sync help center pages, troubleshooting PDFs, screenshots, and escalation notes into one cite-ready support bucket.

pricing

Pricing

Keep packaging docs, plan sheets, procurement FAQs, and discount policies separate from broad product knowledge.

legal

Legal

Give legal-facing agents a governed source set for terms, policies, compliance PDFs, and data handling documents.

customer-onboarding

Onboarding

Assemble setup guides, diagrams, checklists, and customer-specific instructions for onboarding assistants.

REST API

List retrieval buckets

cURL

# List retrieval buckets for this API key
curl https://api.calypso.so/v1/knowledge/buckets \
  -H "Authorization: Bearer sk_..."

{
  "team_id": "team_123",
  "buckets": [
    {
      "id": "bucket_abc",
      "slug": "support-handbook",
      "name": "Support Handbook",
      "status": "active",
      "memberCount": 12,
      "counts": {
        "file": 8,
        "website": 3,
        "qa": 1,
        "retrievable": 12
      },
      "bucketStore": {
        "status": "active",
        "indexed_member_count": 12,
        "pending_member_count": 0
      }
    }
  ]
}

GET/v1/knowledge/buckets

List bucket ids, slugs, source counts, status, and bucket-store readiness for the team behind an API key.

POST/v1/knowledge/files/upload-session

Create a direct upload session and route one file into bucket_ids, bucket_slugs, bucket, or create_missing_buckets.

POST/v1/knowledge/files:batch/upload-session

Prepare 1 to 100 files with shared or per-item bucket destinations, then poll readiness.

Readiness model

Ready means searchable, not just uploaded.

For reliable RAG tests, wait until each source is indexed and the bucket store reports it as active. That is when an agent can retrieve and cite the new knowledge.

Accepted

Part of the durable ingestion path.

Queued

Part of the durable ingestion path.

Indexing

Part of the durable ingestion path.

Active source

Part of the durable ingestion path.

Bucket active

Searchable through the selected bucket.

Get started with Buckets

Create the source layer your agents can trust.

Start with Buckets, attach them to agent profiles, and reuse the same retrieval layer across your website, API, MCP clients, n8n workflows, and product UI.

Get started free Bucket docs