AiMatch

Context Engine

Context Engine gives AI agents and teams shared, persistent project memory across sessions and tools. Teams can preserve decisions, contacts, and files in separate engines and retrieve them through connected AI clients or the app.

One of 672 tools in AI Agents

Visit Context Engine Compare alternatives Own this tool? Claim it
Context Engine screenshot

Who it's for

  • Engineering teams using Claude Code, Cursor, or other supported AI tools that need shared project decisions and conventions across sessions.
  • Consultants or client teams that need separate memory engines and access controls for each client or project.
  • Teams whose people and AI tools need to browse, add, and ask questions about the same project memories.

How it fits your workflow

Create an engine and describe what it is for. Connect AI tools and choose which engines each connection can reach and whether it can read or save.

As work happens, connected tools save memories such as decisions, notes, people, customers, and files. Users can also add files, folders, or ZIP archives in the app.

When a tool starts a session, it receives a brief with recent context and changes since its last visit. It can search for more; people can browse the same memory and ask questions whose answers identify their sources.

Pricing

Free: Free plan costs $0 forever and includes 1 agent per engine, up to 3 when friends join, and 1 GB per engine. No card is required. People are included at no cost per person on every plan.

Free

$0 forever

  • · 1 agent per engine, up to 3 when friends join
  • · 1 GB per engine
  • · No card required

Pro

$12/month or $99/year

  • · 10 agents per engine
  • · 25 GB per engine

Unlimited

$39/month or $299/year

  • · Unlimited agents per engine
  • · 100 GB per engine

Enterprise

From $6,000/year

  • · Runs on your own server and is run by Relote
  • · Single sign-on for your own apps
  • · Updates and support included

Prices checked on Oct 23, 2026 from the vendor's site. They can change; confirm before you buy.

Key features

Shared, persistent memory
AI tools save information as they work and can retrieve it in later sessions, so context is shared across supported tools rather than remaining only in one chat history.
Purpose-specific engines
Create separate memory spaces for a person, client, project, or team. Each engine has a stated purpose that tools use to decide what belongs there.
Per-connection access controls
Choose which engines each connection can reach and whether it can read or read and save. Connections can be paused or cut off.
Memory search and sourced answers
Ask questions in plain language. Answers name the memories they came from, and search can match meaning as well as exact words.
Briefs, summaries, and linked memories
Tools receive a short brief when starting, including recent activity and changes since their last visit. Context Engine also creates daily, weekly, and monthly summaries and links related memories.
Memory history and duplicate handling
Memory changes retain who made them, which tool was involved, and the previous content. When a tool tries to save an existing memory under the same name, it is directed to update that memory.

Works with

  • ChatGPT
  • Claude
  • Claude Code
  • Cursor
  • MCP-compatible tools and agents
  • HTTP clients and custom agents

Limitations to know

  • The Free plan includes 1 agent per engine, up to 3 when friends join, and 1 GB per engine; paid plans specify their own agent and storage limits.
  • When an engine reaches its memory limit, new memories are refused; the page says existing stored memories are not deleted.

Alternatives to consider

  • Atlaso

    Choose it if you want an AI memory layer that explicitly lists a broader set of coding clients, including Codex, OpenCode, and Antigravity.

  • DocuMemory

    Choose it if you want to turn AI chat sessions into concise engineering memories and want a description that explicitly includes self-hosting options.

Compare Context Engine side by side

Getting started

claude mcp add --scope user --transport http ce-invoicing \ https://app.contextengine.com/mcp/c/k7f2q9
  1. Sign in with your email; no password is required.
  2. Create an engine and describe its purpose.
  3. Connect an AI tool and select which engines it can reach and its read or read-and-save access.
  4. For Claude Code, run the displayed MCP command, then enter /mcp, choose the connection, and authenticate.

Frequently asked questions

Is Context Engine a chatbot?

No. The page says it does not run agents; the AI tools do the thinking. Context Engine is shared memory that tools save to as they work and read when they start. People can also ask it questions, and answers name the memories they came from.

Can people and AI tools use the same memory?

Yes. AI tools save and retrieve memories, while people can browse memories, see who saved them, ask questions, read summaries, and add files in the app. The whole team can share an engine.

Which AI tools work with it?

The page lists ChatGPT, Claude on the web and desktop, Claude Code, and Cursor. Other tools that speak MCP or HTTP can connect as well.

Can client memories be kept separate?

Yes. The page says each engine is stored separately and each connection reaches only the engines selected for it, with read or read-and-save permissions.

Can I export my memory?

Yes. An engine can be exported at any time as a ZIP of Markdown and JSON. The page says every memory is a plain Markdown file.

What happens when an engine is full?

The page says users are told before an engine is full. When it is full, a new memory is refused with the reason, and existing stored memories are not deleted.

Where is data stored, and who can access it?

The page says each engine has its own database, search index, and running service on Cloudflare. Support opens an engine only with the user's approval for a specified time, and each visit is logged.