NeatContext Plugins

NeatContext Plugins is an open-source tool by XTSoftwareLabs that turns long AI coding conversations into structured, reusable domain knowledge. It works as a plugin for Claude Code, GitHub Copilot (CLI and VS Code), Kimi Code, Codex, and pi, letting developers save the durable insights from debugging sessions, planning discussions, and incident investigations so they can be reused in future sessions or shared with teammates.

What It Is

NeatContext addresses a specific pain point in AI-assisted development: every time you start a new session with a coding agent, you have to re-explain your team's systems, constraints, terminology, and decisions. NeatContext extracts that domain knowledge from existing conversations and stores it as a structured context — a domain profile plus a knowledge folder — that can be connected to any new session. Instead of carrying over a raw transcript, it keeps only what will help the model work accurately next time.

How the Workflow Works

The core workflow revolves around a small set of slash commands:

• /neatcontext:save [name] — Extracts durable knowledge from the current conversation and saves or updates a named context. Follows familiar Save / Save As behavior.
• /neatcontext:use [name or number] — Connects a saved context to the current session so the coding agent starts with the relevant domain knowledge already loaded.
• /neatcontext:create — Creates a fresh context linked to an existing knowledge folder on disk, without copying or moving files.
• /neatcontext:import / /neatcontext:export — Share context bundles with teammates across machines.
• /neatcontext:mode [auto|ask|manual] — Controls how the session automatically routes to the right context when topic shifts are detected.
• /neatcontext:extensions — Manages MCP server bindings that give a context access to external tools like incident trackers or log stores.

Each Claude Code window maintains its own connected context independently, and saving never overwrites an external knowledge folder you linked manually.

Context Contents and Architecture

A saved context contains two parts: a domain profile (team rules, terminology, constraints, preferred working patterns) and a knowledge folder (TSGs, runbooks, decisions, troubleshooting notes, session summaries). These are generated automatically from the conversation rather than requiring manual authoring.

NeatContext is distinct from simply saving or resuming a conversation. Resuming a conversation gives you the original session history; exporting a conversation gives you the raw transcript. NeatContext gives you a distilled, reusable artifact — the knowledge that will help the model answer accurately in a future session, without the back-and-forth that produced it.

Security and Privacy Model

The README documents a deliberate local-first security model:

• All contexts are stored locally under ~/.neatcontext/contexts.
• The host plugins run only the Node.js files bundled in the repository and make no outbound internet requests.
• Transcript content is read only after an explicit save, through a bundled reader that drops high-risk payloads and applies best-effort secret redaction.
• Extension bindings (MCP server commands, credentials, tokens) live in ~/.neatcontext/extensions.json on the user's machine and are never carried inside a shared context bundle.
• Extension servers are spawned without a shell, receive a minimal environment, and are stopped when the session ends or switches context.

Update: v0.3.4

The latest release is v0.3.4, published on August 10, 2026. The repository was created in late July 2026 and has seen active development, with the most recent push on the same date as the latest release. The project is written in JavaScript, requires Node.js 18 or later, and is licensed under MIT.