Know - Active Memory System

The description states that Know is an Active Memory system for code, implemented as a Rust CLI tool. It stores project rules in versioned TOML files within the repository, and uses a SQLite read model to track rule-to-code relationships. A repository hook injects relevant rules into AI agents (e.g., Codex) before edits are made, helping prevent unintended consequences of code changes.

• The system supports:
  • Rule constraints with rationale
  • Links to files or globs it protects
  • Verification state tracking
  • Advisory behavior: rules are surfaced but not enforced automatically

Inference: This is a developer tool aimed at improving AI-assisted coding workflows by providing context-aware constraint information.

The author positions Know as a solution to the problem of implicit knowledge in codebases, where important architectural decisions or business rules live only in senior engineers' heads or stale documentation. The system aims to make these constraints visible before changes are made, rather than discovering them after a regression.

• Claims:
  • "Know provides rules, decisions, and constraints that apply to code before a human or AI agent changes it."
  • "It makes those constraints visible before the change, rather than discoverable only after the regression."

Inference: The positioning is focused on developer experience, AI agent alignment, and risk mitigation in codebases. It does not claim market traction, revenue, or adoption beyond its own demo.

The description states that Know targets developers working with AI agents (like Codex) who are about to edit code. The system is designed to help prevent unsafe refactors by surfacing constraints at the point of change.

• It supports:
  • AI-assisted coding workflows
  • Developers who want to avoid breaking business logic or architectural decisions

Inference: The primary customer is likely technical teams using AI agents for code modification, especially those with large, complex codebases where implicit knowledge causes issues.

Not evidenced.

The description does not mention any pricing model, monetization strategy, or business model.

• Built in Rust CLI
• Uses SQLite for read model
• Stores rules in TOML files
• Integrates with Codex via repository hooks
• Includes:
  • Automated tests
  • Checksum-verified installer
  • Demo script (./demo.sh)
  • Fast path installation: git clone, ./install.sh, ./demo.sh

Inference: The tool is built for developers and integrates into existing workflows. It’s designed to be lightweight, reproducible, and focused on immediate utility.

Not evidenced.

There is no mention of:

• Customers
• Revenue
• Usage metrics
• Adoption beyond the author's own demo
• Product iteration or feedback loops

The project is described as an MVP built in a 90-minute hackathon, with no indication of ongoing development or user base.

Not evidenced.

No comparison to existing tools or platforms is provided in the description.

• No evidence of real-world usage or adoption — all activity appears limited to a demo.
• Single-person team — only one member listed (Emanuel Lindström).
• Self-reported, unverified claims — no external validation or third-party data.
• Limited scope — the MVP is explicitly described as small and focused on core functionality.
• No pricing or monetization strategy — unclear if this will ever become a commercial product.

1. Has Know been tested in any real-world codebases beyond the demo?
2. What are the plans for expanding beyond the current CLI and repository hook integration?
3. Are there any partnerships or early adopters already using Know?
4. How does Know handle scalability or performance with large repositories?
5. Is there a plan to support IDE integrations (e.g., VS Code, JetBrains)?
6. What is the long-term vision for Know beyond the current MVP?