Model Context Protocol

Let your AI
rotate your keys.

Detect. Rotate. Fix — automatically.

Your AI agent finds leaked credentials, rotates them, and verifies the fix before you even know there was a problem.

Most tools detect problems. Lockzero fixes them.

Autonomous — no human requiredEvery action fully auditedConnect in 30 seconds
Live Incident
Responding
Found OpenAI key in CI logs
assessing blast radius...
rotating credential...
verifying adoption...
Secret secured. No downtime.
Time to remediation4.2s
Human requirednone
Powered by Lockzero MCP · full audit logged
Autonomous

By default, your agents respond in seconds, not minutes.

Detection in real-time

An AI agent detects the leak, calls the MCP endpoint, and initiates remediation before a human is even aware there was a problem.

Rotation without humans

Lockzero validates the new key, transitions traffic, and retires the old credential. The whole cycle completes programmatically, end to end.

Full audit trail regardless

Every MCP-triggered action is logged with the same fidelity as human operations — actor, timestamp, outcome, and affected secret.

The Difference

Other tools stop at the alert.

Most tools
detect
alert
wait for a human
Lockzero MCP
detect
rotate
verify
done

Credentials managed at machine speed. No humans required. Full audit preserved.

12 MCP Methods

Everything your AI agent can do with Lockzero

Each capability is a discrete MCP method, callable via JSON-RPC 2.0. All actions are tenant-scoped, policy-checked, and logged.

detect_exposure

Detect leaked credentials

Scan a text payload — logs, diffs, error output — for exposed API keys. Returns matched secret IDs, confidence scores, and recommended action.

list_secrets

List managed secrets

Return all secrets in the tenant with status, rotation age, provider, and environment labels. Useful for pre-deploy audits and status summaries.

get_secret_metadata

Query secret metadata

Get full metadata for a specific secret: provider type, environment, last rotation timestamp, rotation schedule, rotation policy, and current status.

assess_blast_radius

Assess blast radius

Map which services, environments, and downstream systems are using a given credential. Understand exposure scope before triggering remediation.

rotate_secret

Rotate a credential

Trigger the full rotation workflow for a secret. Lockzero validates the new key, transitions traffic, and retires the old credential. Returns new ARN and status.

rollback_secret

Rollback to previous key

Instantly revert a secret to its previous working version. Designed for post-rotation breakage. Returns the restored secret ARN and confirmation.

revoke_secret

Emergency revocation

Immediately invalidate a credential when exposure is confirmed. Marks the secret as revoked in the audit trail and triggers downstream alerts.

verify_remediation

Verify remediation

Confirm that a rotated credential is active, the old key is fully retired, and all dependent systems have adopted the new value.

get_rotation_history

Get rotation history

Retrieve the full rotation history for a secret: timestamps, actors (human or agent), outcomes, and rollbacks. Useful for compliance and incident reports.

get_active_incidents

Pull active incidents

List all open security incidents in the tenant — leaked keys, overdue rotations, policy violations, and usage anomalies. Scoped to your account.

check_rotation_policy

Check rotation policy

Query the rotation policy for a given secret: max age, required approvals, notification rules, and whether the secret is overdue.

get_audit_log

Query the audit log

Retrieve recent audit events for a specific secret or the full tenant. Includes human and agent-triggered actions, with actor ID and outcome.

Autonomous Workflow

From leak detection to verified remediation

The full MCP workflow runs end-to-end without human intervention. Every step is logged and auditable.

1

Leak detected

Agent finds a credential in logs, a diff, an error response, or a notification — and calls detect_exposure.

2

Scope assessed

Agent calls assess_blast_radius to understand which services are at risk before taking any action.

3

Rotation triggered

Agent calls rotate_secret. Lockzero validates the new key, transitions traffic, and retires the old credential.

4

Remediation verified

Agent calls verify_remediation to confirm the old key is retired and all systems have adopted the new credential.

Use Cases

What teams are building with Lockzero MCP

Automated incident response

Connect your alerting system to an AI agent. When an alert fires for unusual API usage, the agent calls Lockzero MCP to assess the affected secret, rotate it, and close the incident — all before an engineer is even paged.

Pre-deploy security checks

Before every production release, a CI/CD agent calls list_secrets and get_active_incidents to verify no credentials are overdue for rotation or involved in open incidents. Block deployments when security posture is degraded.

Continuous compliance monitoring

A scheduled agent polls check_rotation_policy and get_rotation_history to build a running compliance report. Flag secrets that have exceeded their max rotation age without requiring any human to run a report manually.

Post-leak forensics

After a confirmed credential exposure, an agent calls assess_blast_radius, get_rotation_history, and get_audit_log to build a complete timeline — which systems were affected, when the key was last rotated, and what actions were taken.

Setup

Connect in 30 seconds

No SDK required. Lockzero MCP speaks standard JSON-RPC 2.0 over HTTPS — any MCP-compatible client works out of the box.

1

Subscribe + generate your MCP key

After subscribing, open your Lockzero dashboard → Settings → MCP Keys. Generate a new key. It starts with lz_mcp_ and is shown once — copy it now.

2

Add Lockzero to your AI client config

Paste the endpoint and your key into your client config (Claude Desktop, Claude Code, or any MCP-compatible client). See exact snippets below.

3

Tell your agent to manage secrets

Say "Check my secrets status" or "Rotate the OpenAI key if it's overdue." Your agent now has full programmatic access to your Lockzero account.

~/Library/.../claude_desktop_config.json
// Claude Desktop — add to your config file{
  "mcpServers": {
    "lockzero": {
      "url": "https://lockzero.io/api/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_MCP_KEY"
      }
    }
  }
}
terminal
# Claude Code CLI — run once in your terminalclaude mcp add lockzero \
  --transport http \
  https://lockzero.io/api/mcp \
  --header "Authorization: Bearer YOUR_MCP_KEY"

# Verify it's connected
claude mcp list
Technical Reference

Integration details

Endpoint & protocol

Endpointhttps://lockzero.io/api/mcp
ProtocolJSON-RPC 2.0 over HTTPS POST
Content-Typeapplication/json
Auth headerAuthorization: Bearer <mcp-key>
Response formatJSON-RPC 2.0 result or error object
POST /api/mcp
POST https://lockzero.io/api/mcp
Authorization: Bearer <mcp-key>
Content-Type: application/json

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "rotate_secret",
  "params": {
    "secret_id": "prod/openai/api-key",
    "reason": "Exposure detected in CI logs"
  }
}

Tenant isolation

Each MCP key is scoped to a single Lockzero tenant. Agents can only access secrets and incidents within the account the key belongs to.

Principle of least privilege

MCP keys can be issued with read-only or read-write permissions. Grant agents only the access level they need for their specific workflow.

Full auditability

Every MCP-triggered action is logged identically to human operations — method, actor, timestamp, affected secret ID, and outcome.

Compatibility

Works with any MCP-compatible client

Lockzero MCP follows the Model Context Protocol standard. Any agent or AI platform that supports MCP can connect and use the full capability set.

Claude (claude.ai)Claude CodeCustom MCP clientsAny JSON-RPC 2.0 client

Give your AI control over your credentials.

Fix incidents before you see them. Credentials managed at machine speed, with full audit preserved.

Connect your first agent in under 2 minutes
Tenant-scoped · Policy-enforced · Fully audited · Actor + timestamp on every action
Lockzero MCP — Let Your AI Rotate Your Keys | Lockzero