DOCS
What it takes to run PathWise.
This page covers what ships today: what you need, how it installs, how you use it, and how to write your own module. Where something is planned, it says so.
Scoped to the current release. Planned features are tagged.
What you need.
Install.
After you buy, you get a signed download link and a license file by email. The engine is one Docker image. Your license is a small signed file that rides alongside it.
- Download
Open the signed link from your email and download two things: the engine image and your license.json.
- Load the image
Run docker load to bring the engine image into Docker on your machine.
- Add your license
Mount your license.json when you run the engine. The license check is soft, so it never blocks you, even fully offline.
- Set your model key
Provide your model API key as an environment variable. The engine reads it at startup and stops with a clear message if it is missing.
The exact commands ship with your download, written for your machine. Delivery and the buyer download flow are part of the launch build.
Configure.
PathWise reads a few settings from the environment. These are the ones that matter today.
The two official modules, DNS and Check Point, are bundled and load by default.
How you use it.
You paste a ticket, the kind you already get: a user cannot reach an app, a login is refused, a name will not resolve. PathWise picks the entry layer from the error and walks the access path, reading the evidence your modules expose and running its own read-only network probes.
It returns a structured answer: the finding, the evidence behind it, a confidence level, and an honest note on what it could not verify. Then it stops at a proposed solution for you to perform. It never changes anything on your systems.
Where the model runs is your choice.
PathWise sends its reasoning to a model. You decide where that model lives. One placement works today, two more are planned.
The model runs on the provider's cloud, reached over its public API with your key. Anthropic today.
The model runs in your own cloud account, on Bedrock or Azure OpenAI, so the data stays inside your cloud boundary.
The model runs on your own server through Ollama or vLLM, so data never leaves the box.
Today the reasoning runs against the Anthropic API. The in-tenant and local placements arrive with the settings work. We tag them planned so you are not surprised.
Write a module.
The Open Module Standard is free and public. A module is a small folder of plain files. You can teach PathWise about a tool it does not ship with.
- Make a folder
Create a folder in your modules directory named for the tool.
- Write module.yml
Fill in the manifest: the name, the vendor, the version, and which of the eight layers the module reads.
- Write knowledge.md
Describe the tool in plain language: what it checks, what healthy looks like, what broken looks like, and the common causes.
- Choose a door
Point the module at its evidence. File drop works today: export to CSV or JSON and the engine reads the files. Live and custom connections are in the standard and arrive later.