Quickstart

Install Velixar

Governed, auditable memory for your AI — running in four steps. Every snippet below is executed against the live API before it ships to this page.

  1. 1

    Install the SDK

    Get started by installing the Velixar package. Python 3.9+ or Node 18+.

    python
    pip install velixar
  2. 2

    Initialize the client

    Create a key in the dashboard (Setup → Get an API key) and choose its scopes: read, write, and optionally delete.

    python
    from velixar import Velixar

client = Velixar(api_key="your-api-key")

client.health()
# {'status': 'ok', 'encryption_at_rest': 'enforced', 'audit_persistence': 'durable'}
  3. 3

    Add memory

    Store what matters about each end user. One API key, many users — pass user_id and the memory belongs to them.

    python
    client.store(
    "Alex is a vegetarian and is allergic to nuts.",
    user_id="alex",
    tags=["diet", "allergy"],
)
# -> '6024acec-6533-4fef-bc2c-a32e2cdf50cb'
  4. 4

    Retrieve memory

    Search by meaning, scoped to one user. Alex's results never contain another user's memories.

    python
    results = client.search(
    "What can I cook for dinner tonight?",
    user_id="alex",
)

for m in results.memories:
    print(m.content, m.score)
# Alex is a vegetarian and is allergic to nuts.  0.3644

Using an agent instead?

Claude Code, Cursor, Windsurf and any MCP host can talk to Velixar with no code at all — 37 tools and 19 workflow playbooks, over stdio.

claude code
claude mcp add velixar \
  --env VELIXAR_API_KEY=your-api-key \
  -- npx -y velixar-mcp-server@1.5.0

Every call — store, search, delete — is written to an immutable, tamper-evident audit trail. Memory content is encrypted at rest.

Give your AI a memory you can defend.

Free tier, no card. Encrypted at rest, tamper-evident audit trail, and every memory carries its origin.

Get an API key →Read the docs