PathruleWhy a Context Layer Should Have an Open Core
A team should be able to inspect its knowledge and the integrations that deliver it.
An open foundation makes that possible, even when advanced intelligence and managed services are separate products.
What this covers
- Pathrule’s foundation, the CLI, the local knowledge store, the hooks, the MCP server and the pattern bundles, is open source; Pathrule Cloud is proprietary.
- Local knowledge can be used without an account. The optional model is a separate capability.
- Readable knowledge and client configuration let teams review what agents receive without promising that every selection algorithm is open.
Comparison
| Question | Open foundation | Intelligence and Cloud |
|---|---|---|
| What it does | Local knowledge storage, CLI, MCP and client integration | Task-scoped intelligence and managed workspace services |
| Where it runs | Local mode on your machine, no account required | Intelligence capabilities locally where supported; Cloud as a managed service |
| Who can read it | Public source repositories | Knowledge and delivered output are inspectable; the engine and service source are proprietary |
| What it costs | Open foundation under Apache-2.0 | Availability and limits follow Pricing |
| How you change it | Fork the public source or send a patch | Edit your knowledge and integrations; request product changes |
A foundation you can inspect
A context layer sits close to work. It stores the decisions, constraints and procedures an agent may rely on, so those records should be readable and editable by the people who own them.
The store and the integration are useful things to inspect. You can see which workspace is attached, which client files Pathrule wrote and which tools the MCP server exposes.
The source boundary should be plain
Pathrule’s CLI, local knowledge foundation and patterns are open source. Local mode needs no account and stores your knowledge on your machine.
Pathrule Cloud is proprietary: managed accounts, organizations, workspaces, persistence and access, Remote MCP, and the Cloud model tier. It builds on the open foundation rather than replacing it. See Open source core for the repositories and licenses.
Local knowledge remains useful offline
A local workspace can answer knowledge requests without a Cloud account, and nothing on the machine runs a model: what Pathrule learns locally, it learns from your own words and the shape of your commands.
Cloud sharing is a separate choice. A teammate can use the same workspace knowledge from another machine, while each person keeps their own repository attachment and client setup.
Review the knowledge and what was delivered
An open foundation does not mean every selection algorithm is public. It means the source boundary is stated clearly and the knowledge you depend on remains yours to inspect.
Open the records, inspect the generated client files and check the context your agent received. These are concrete ways to review the layer that sits between team knowledge and generated work.
Build on the foundation
The public repositories are listed at github.com/pathrule. Contributions to the open foundation can improve local workflows for everyone.
Questions about the product boundary or a local integration can go to hello@pathrule.io.