Run CortexDB locally

The one-command path

Multi-arch image on Docker Hub — linux/amd64 and linux/arm64. Data persists to the named volume.

docker run -d \
  --name cortexdb \
  -p 3141:3141 \
  -v cortexdb-data:/data \
  -e OPENAI_API_KEY=$OPENAI_API_KEY \
  -e ANTHROPIC_API_KEY=$ANTHROPIC_API_KEY \
  cortexdb/cortexdb:latest

Once it is up, the dashboard is at localhost:3141 and health is at localhost:3141/v1/admin/health.

Your data is stored locally, but the default configuration still calls out for model work: OpenAI for embeddings and entity extraction, Anthropic for answer generation. Consolidation runs on the box. To drop the embedding dependency too, point CORTEX_EMBEDDING_URL at a local Ollama — the profiles doc puts the cost at roughly 5pp of recall.

No Docker required

A single stripped executable plus bundled assets, README and docs. Untar and run.

tar -xzf cortexdb-0.9.8-linux-amd64.tar.gz
cd cortexdb-0.9.8-linux-amd64
./cortexdb

SHA-256 checksums are attached to every GitHub Release. macOS and Windows builds are not shipped yet — ask us if you need one.

Write one thing, then ask for it back

Every write is an event with a time it was observed and a key that makes it idempotent. Nothing is overwritten.

curl -X POST http://localhost:3141/v1/experience \
  -H 'Content-Type: application/json' \
  -d '{
    "scope": "ws:demo",
    "modality": "observation",
    "content": { "kind": "text", "text": "Priya at Acme signed for 200 seats." },
    "context": { "observed_at": "'`date -u +%Y-%m-%dT%H:%M:%SZ`'" },
    "idempotency_key": "'`uuidgen`'"
  }'

Recall returns a pack, not a list of chunks: events, episodes, facts and beliefs come back together in one round trip, already assembled into a context block.

from cortexdb.v1 import V1Client

client = V1Client(api_url="http://localhost:3141", actor="agent:demo")

pack = client.recall(
    scope="ws:demo",
    query="What did Priya commit to?",
    view="holistic",
    include=["events", "episodes", "facts", "beliefs"],
    budgets={"max_tokens": 1200},
)

print(pack["context_block"])

pip install cortexdbai or npm i cortexdbai — both ship the same client. Full endpoint reference in the docs.

What lands on disk

Bundled in the tarball, and in the image at /opt/cortexdb/docs/.

• cortexdbstripped release binary, ~80 MB
• assets/WordNet synonym map for BM25 expansion
• docs/GETTING_STARTED.mdfive-minute quickstart
• docs/API_DESIGN_V1.mdendpoint design rationale
• docs/API_REFERENCE_V1.mdfull endpoint reference
• docs/CORTEXDB_FOR_THE_ENTERPRISE.mdsecurity, compliance, deployment
• docs/cortexdb_customer_deck.htmlproduct overview deck
• docs/PRODUCTION_DEPLOYMENT.mdsystemd, reverse proxy, secrets
• README.md, LICENSE.txtterms of the source-available licence