Knowl

Persistent memory layer for AI coding agents with provenance tracking

Coding & Development AI Models & APIsSubscription

Visit website ↗

$20Starting price

1.4KMonthly visits

2Popularity

About Knowl

Knowl provides a persistent memory system for AI coding agents that records the origin, review trail, and successor relationships of stored facts. Each piece of knowledge includes reasoning, evidence, and provenance, allowing agents to return current answers while enabling users to verify how decisions were made. The system runs locally within a repository and maintains a shared memory that survives agent sessions. Facts are never deleted but are superseded with successor records, preserving historical context indefinitely. Knowl operates as an MCP server, integrating with various AI coding tools through adapters or direct MCP support. The tool emphasizes checkable knowledge by design, ensuring that claims can be validated without relying on human recall. It is designed to prevent knowledge staleness by tracking changes and maintaining a clear audit trail of decisions over time.

Key features

Provenance tracking for every fact
Successor relationships to track changes over time
Local-first operation with optional sharing
MCP server compatibility
Review trail and conflict governance
Audit trail for decision tracking
Workspace-based memory sharing
Supports multiple AI coding tools

Use cases

Maintaining consistent agent memory across sessions
Tracking decisions and their evolution over time
Validating agent outputs against documented reasoning

Pros

Records provenance, reasoning, and evidence for every stored fact
Supports local operation with optional sharing within workspaces
Tracks successor relationships to maintain historical context
Integrates with multiple AI coding tools via MCP adapters
Provides checkable knowledge without requiring human intervention

Cons

No free tier beyond the initial trial period
Limited to Node.js 20 or newer for installation
Requires MCP-compatible tools for full functionality
Pricing is organization-based, not per workspace

Frequently asked questions about Knowl

What does Knowl do?

Knowl provides a persistent memory system for AI coding agents that records the origin, review trail, and successor relationships of stored facts. Each fact includes reasoning, evidence, and provenance, allowing agents to return current answers while enabling users to verify how decisions were made.

Who is Knowl designed for?

Knowl is designed for developers and teams using AI coding agents who need to maintain accurate, verifiable knowledge over time. It is particularly useful for organizations where knowledge staleness or lack of provenance can lead to errors or inefficiencies.

How does Knowl integrate with AI coding tools?

Knowl operates as an MCP server and integrates with various AI coding tools through adapters or direct MCP support. It offers one-command setups for tools like Claude Code, Cursor, GitHub Copilot, and others, as well as support for custom MCP-compatible tools.

Does Knowl store data externally?

No, Knowl runs locally within a repository and keeps its knowledge there. Code, embeddings, and session history are never uploaded externally, ensuring data remains under the user's control.

Can Knowl handle changes to facts over time?

Yes, Knowl is designed to handle changes by recording successor relationships. When a fact is updated, the new version supersedes the old one, but the old version is never deleted, preserving historical context indefinitely.

How do I get started with Knowl?

To get started, install Knowl via npm, initialize it in your repository, and connect it to your preferred AI coding tool using the provided commands. Detailed setup instructions are available in the documentation.