Skip to main content

Getting Started

This guide takes you from a fresh install to your first grounded answer. It follows the same shape as a real workspace: connect a model, add an extension, then add a domain profile and a knowledge base.

Prerequisites​

  • The NeatContext desktop app — installed and able to open.
  • Your own LLM provider — an OpenAI-compatible or Anthropic-compatible API key/endpoint, or a local model. NeatContext does not host a model; it orchestrates yours. A tool-calling-capable model is required to use extension tools.
  • (Optional) Node.js 18+ — only if you plan to run local extensions or the incident demo.

Step 1 — Configure your model​

  1. Launch NeatContext.
  2. Open model settings and add your provider: base URL, API key, and model name.
  3. Choose a tool-calling-capable model and make it the active model. Your active model is shown in the top bar.
info

No inference happens on NeatContext's servers — requests go directly from the app to the provider you configured.

Step 2 — Add an extension (optional but recommended)​

Extensions give the model tools for your real systems. To add one:

  1. Go to the Extensions page.
  2. Click Add extension and select the extension's folder — the one containing its neatcontext-extension.json manifest. NeatContext copies it into its own userData/extensions/ and loads it.
  3. Enable the extension. If it declares connection: none, there is nothing to authenticate. If it requires a connection, complete the connection step so its tools become available.

Once enabled, the extension's tools are offered to the model during chat. You can try a ready-made connector by following the Incident Analysis walkthrough, or write your own with the extension guide.

Step 3 — Add a domain profile​

A domain profile steers the model toward your team's correct behavior.

  1. In the Domain Profiles panel, click Import local Markdown profile and choose your profile .md file.
  2. Mark it active.

If you don't have a profile yet, start from this minimal template and save it as my-team.md:

---
id: my-team
name: My Team
type: team
owner: My Team
---

# My Team

## What we own
- (services / systems your team is responsible for)

## First checks during an incident
1. (the first thing your team looks at)
2. …

## Dangerous actions (do NOT do without approval)
- (irreversible or high-blast-radius actions to avoid)

## Response style
- Separate facts from hypotheses, and cite the runbook you relied on.

Step 4 — Add a knowledge base​

Add a folder of your team's Markdown docs (runbooks, TSGs, postmortems) as a local knowledge folder. NeatContext searches it to ground answers in your material. You can add more than one folder, and remove folders you don't want the model to search.

Step 5 — Ask your first question​

Open a new chat and ask something your profile and knowledge base can answer. For an operational workspace, that might be:

Please analyze this incident: <link or ID>.
What should we check first, and what's the safe action?

With a tool-calling model and an enabled extension, the model will call tools to gather first-hand evidence, search your knowledge base, and answer within your profile's guardrails — citing the documents it used.

Next steps​