KnownsDocuments
Contributing

Philosophy

Core principles behind Knowns, based on the main GitHub repository philosophy document

Philosophy

This page summarizes the core philosophy from the Knowns repository.

Knowns is built around a simple idea: AI should understand your project the same way a teammate does.

AI Should Read, Not Guess

Knowns is designed so AI reads real project artifacts instead of depending on prompt tricks or repeated explanations.

That means:

  • context should be explicit, structured, and versioned
  • important knowledge should not live only in chat history
  • references should resolve back to real project files

Files Are The Source Of Truth

Knowns is file-first and local-first.

  • tasks are files
  • docs are files
  • structure lives on disk
  • project knowledge stays versionable in Git

This avoids hidden state and keeps knowledge portable.

Explicit Context Beats Memory

Knowns favors resolvable references such as @doc/... and @task-... over fuzzy memory.

The goal is deterministic context loading instead of guesswork.

Local-First, Sync Optional

The local project remains the source of truth.

If teams use sync or shared visibility features later, those layers should not replace local files.

CLI First

Knowns is built for developers who need a scriptable, composable, automation-friendly interface.

The Web UI exists to visualize and browse, not to replace the CLI.

AI As Teammate

Knowns treats AI like a teammate that should:

  • read tasks
  • read documentation
  • follow references
  • execute structured instructions consistently

Keep The Model Simple

Knowns intentionally centers a few primitives:

  • tasks
  • documents
  • references
  • files

This keeps the system easier to reason about and harder to break.

What Knowns Is Not

To stay focused, Knowns avoids becoming:

  • a full project management suite
  • a ticketing platform
  • a SaaS dashboard
  • a replacement for Git
  • a generic note-taking app

Core Belief

If knowledge is explicit, structured, and versioned, AI can work like a real teammate.

  • Contributing - How to contribute within these principles
  • Roadmap - Current direction from the main repository
  • License - MIT license summary

On this page