Agent Secret Store DocsSign up
Documentation — v1

Agent Secret Store

The secure credential vault purpose-built for AI agents. Store API keys, tokens, and secrets — then issue scoped, short-lived access tokens to your agents instead of raw credentials.

Zero to vault in 30 seconds

quickstart.py
from agentsecretstore import AgentVault

vault = AgentVault(agent_key="your-agent-key")

# Store a secret
await vault.set_secret("production/openai/api-key", "sk-...")

# Retrieve it later
secret = await vault.get_secret("production/openai/api-key")
print(secret.value)  # "sk-..."

What just happened?

Your secret was encrypted with AES-256-GCM using a per-secret data encryption key (DEK), then the DEK was wrapped by a per-tenant key encryption key (KEK) stored in GCP KMS. Your raw secret never leaves your trust boundary unencrypted.

Explore the docs

Why not just use .env files?

Problem with .envAgent Secret Store solution
Secrets baked into container images / VMsSecrets fetched at runtime, never stored in code
No access control — any code reads all secretsScoped tokens limit each agent to exactly what it needs
No visibility into who accessed whatEvery access logged in immutable audit trail
Secret rotation requires redeploymentRotate a secret; all future tokens get the new value automatically
No human oversight for critical credentialsApproval workflows gate access to sensitive secrets
Plaintext at rest if disk is compromisedAES-256-GCM + KMS envelope encryption

Integration guides