Skip to main content
πŸ“– The AI Tool Bible
Context Data preview image
Context Data logo

Context Data

βœ“ Text source-checked Nov 10, 2026

InterLock is a proxy between AI agents and databases, files, and APIs.

Source-checked against the vendor's site Β· rebuilt from the vendor's own pagesProfile updated Link checked How we rate

FreeRAGMulti-model (not source-checked)6.8 / 10

In short

InterLock is an open-source proxy that secures AI agent access to databases and APIs by enforcing role-based policies, redacting sensitive data, and auditing all activity without requiring code changes.

InterLock is a proxy that sits between AI agents and data stores such as databases, files, and APIs. The vendor says that the system authorizes requests by role and policy, redacts sensitive fields, holds risky writes for approval, and audits all activity. Existing agents and data stores do not require changes to integrate with InterLock.

Each agent is assigned a unique identity and key for access control. Roles allow or deny actions per source, with granular control down to the table and column level for SQL sources. Personal data in query results is automatically detected and masked before reaching the agent. The vendor says that all write operations are classified by risk level before execution. The vendor says that every request, whether allowed or denied, is recorded in an append-only audit log containing identity, source, decision, and redaction details.

InterLock supports the Model Context Protocol, Postgres wire protocol, and HTTP. Integration requires only a connection-string change rather than code modifications. The product deploys as infrastructure within the user's own environment.

InterLock is open source under the Apache 2.0 license.

From the vendor's pages

According to Context Data: InterLock is a proxy that sits between AI agents and data stores such as databases, Files, and APIs. Its pages list this strength: Granular access control down to the column level for SQL sources.

β€” The AI Tool Bible editorial team

Pros

  • βœ… Granular access control down to the column level for SQL sources.
  • βœ… Automatic detection and masking of personal data in responses.
  • βœ… Risk-based classification of write operations before execution.
  • βœ… The vendor says that comprehensive append-only audit logging for all requests.
  • βœ… Simple integration via connection-string change without code modifications.
  • βœ… Open source availability under the Apache 2.0 license.

Facts

Platforms
api (not source-checked)
API available
Yes (not source-checked)
Open source
Yes
Model used
Multi-model (not source-checked)
Hacker News
1 story Β· top 1 points Β· last May 16, 2024 Β· checked Nov 10, 2026
Earliest Wayback capture
2024-01-12 (earliest archived capture of the site, not a launch date)

Where this tool sits

Use cases

governed-access-production-datasensitive-data-redactionrisky-write-approval

Frequently asked

How much does InterLock cost?
InterLock is free and open source under the Apache 2.0 license. The provided documentation does not list specific pricing tiers or additional costs for deployment or support.
Do I need to modify my existing code to use InterLock?
No, existing agents and data stores do not require changes. Integration requires only a connection-string change rather than code modifications, supporting the Model Context Protocol, Postgres wire protocol, and HTTP.
How does InterLock handle sensitive data in query results?
Personal data in query results is automatically detected and masked before reaching the agent. This redaction ensures sensitive fields are protected while maintaining the flow of necessary information to the AI agent.
Can InterLock control write operations for AI agents?
Yes, all write operations are classified by risk level before execution. Risky writes are held for approval, allowing organizations to govern and monitor potentially dangerous actions taken by AI agents.
What kind of access control does InterLock provide?
Each agent has a unique identity and key. Roles allow or deny actions per source with granular control down to the table and column level for SQL sources, ensuring precise authorization.
Is there an audit log for all activities?
Yes, every request, whether allowed or denied, is recorded in an append-only audit log. This log contains identity, source, decision, and redaction details for comprehensive tracking and compliance.
See alternatives to Context Data β†’

Explore related

Reviews