> ## Documentation Index
> Fetch the complete documentation index at: https://docs.ampup.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# AmpUp Memory

> Memory for AI agents that learns advice from the runs that went well and serves it before the next one: hosted at memory.ampup.ai, or open source to run yourself

AmpUp Memory is the memory for agents that gets better with use. A conversation is a session; each task in it
asks for advice and gets one skill for that kind of task: a note learned from earlier rewarded runs or, before any
exist, a procedure compiled from your documents. When the session ends, its transcript and outcome are judged or
rewarded, and a rewarded session teaches the next one.

Underneath is a cited memory: notes, documents and conversations become searchable passages and facts, each fact
quoting its source and visible only to people who may read every source it cites. Everything is stored as Apache
Iceberg tables in a catalog and bucket you run.

> [card] [GitHub](https://github.com/A79-ai/ampup-memory)
>
> Source code, Apache-2.0. Issues, discussions and contributions.

> [card] [Why it learns](https://www.ampup.ai/ampup-memory)
>
> How the learning loop works, and the benchmarks behind it.

## Start hosted

AmpUp runs it for you at [memory.ampup.ai](https://memory.ampup.ai/?utm_source=docs\&utm_medium=docs\&utm_content=overview). Sign in with Google and press **Create my
memory**: you get your own org, with you as its admin, free on the Starter plan for 30 days (1,000 episodes and 2M
model tokens a month). The page that follows shows your personal key once and the install line to copy:

```text theme={null}
/plugin marketplace add A79-ai/ampup-memory
/plugin install ampup-memory@ampup-memory
/ampup-memory:login https://memory-api.ampup.ai
```

Typed in Claude Code, that adds the plugin and signs you in through the browser. Other agents connect to
`https://memory-api.ampup.ai/mcp`; the [quickstart](https://docs.ampup.ai/ampup-memory/quickstart.md#hosted) has
the command-line install and what the key is for.

## Connect your agents

> [card] [Claude Code](https://docs.ampup.ai/ampup-memory/coding-agents.md)
>
> A plugin with hooks: every session asks for advice and records how it went.

> [card] [Codex, Cursor, Gemini CLI, Copilot](https://docs.ampup.ai/ampup-memory/coding-agents.md#codex-gemini-cli-cursor-and-copilot)
>
> The same loop through `ampup-memory hooks install`.

> [card] [Any MCP client](https://docs.ampup.ai/ampup-memory/mcp.md)
>
> Search, remember and get advice over MCP at `/mcp`.

> [card] [Your own agent](#from-code)
>
> Python and TypeScript clients, or plain HTTP.

## Run it locally

Install the server and CLI from the public releases, then start a local server and point Claude Code at it:

```bash theme={null}
curl -fsSL https://raw.githubusercontent.com/A79-ai/ampup-memory/main/install.sh | sh

OPENROUTER_API_KEY=... AMPUP_MEMORY_EXTRACT_GOAL_BLIND=1 ampup-memory serve --local   # http://127.0.0.1:8080
claude mcp add --transport http ampup-memory http://localhost:8080/mcp
```

`--local` needs no Docker, catalog service or sign-in, and binds to 127.0.0.1. Without a model key it stores and
searches but learns nothing; any OpenAI-compatible, Anthropic, Bedrock or Vertex AI endpoint, or a local Ollama,
works instead of OpenRouter ([configuration](https://docs.ampup.ai/ampup-memory/configuration.md)). The
[quickstart](https://docs.ampup.ai/ampup-memory/quickstart.md) walks through a first note, search and piece of advice, including fully
offline on a laptop, and building from source.

## From code

```bash theme={null}
pip install ampup-memory        # Python 3.10+
npm install ampup-memory        # Node 20+
```

```python theme={null}
from ampup_memory import AmpUpMemory

mem = AmpUpMemory("http://127.0.0.1:8080", None)
mem.remember("note:deploy", "Deploy billing with ./deploy.sh billing.")
print(mem.query({"find": "passages", "text": "deploy billing"})["results"])

run = mem.session("deploy the billing service")  # advice per task, rewarded at the end
print(run.opening["advice"])
run.end("deployed", reward=1)
```

The clients wrap the HTTP routes: `/remember`, `/search`, `/query`, `/advice`, `/sessions`, `/sessions/end`,
`/runs`, `/runs/reward` and `/mcp` ([all routes](https://docs.ampup.ai/ampup-memory/http-api.md)).

## Your team's hosted memory

To start one yourself, see [Start hosted](#start-hosted). If AmpUp set up a memory for your company, your admin gets
an invite email to set up the org; see [Set up your org](https://docs.ampup.ai/ampup-memory/admin-setup.md) and
[Invite people](https://docs.ampup.ai/ampup-memory/invites.md).

## Guides

> [card] [Concepts](https://docs.ampup.ai/ampup-memory/concepts.md)
>
> Scopes, episodes, facts, branches, sessions and advice.

> [card] [Advice](https://docs.ampup.ai/ampup-memory/advice.md)
>
> How notes are learned, judged and gated.

> [card] [Schemas and queries](https://docs.ampup.ai/ampup-memory/schemas-and-queries.md)
>
> Typed memory, extraction and search ranking.

> [card] [Self-hosting](https://docs.ampup.ai/ampup-memory/self-hosting.md)
>
> Iceberg REST (Polaris), BigLake, Snowflake, Databricks; GCS, S3, Azure.

> [card] [CLI](https://docs.ampup.ai/ampup-memory/cli.md)
>
> The `ampup-memory` and `ampup-memory-client` commands.

> [card] [Security model](https://docs.ampup.ai/ampup-memory/security-model.md)
>
> Identity, ACLs, API keys, audit, deletion and retention.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.