# Oort

> Oort is open-source long-term memory for AI agents. The `oortmem` 0.9.0 package is the TypeScript SDK and the `oort` CLI on npm, and the native Python client on PyPI. These files teach a coding agent to build Oort into your own apps.

Start with the oort-sdk skill (SKILL.md): it picks the integration, shows the agent loop and states the rules every integration follows, then links the reference for each task. The skill, its references and the three Docs files are also in one file, [llms-full.txt](https://oortcloud.dev/llms-full.txt) (217 KB). Source: oortmem/oort v0.9.0, oortmem 0.9.0, commit a25598daeb1ef5f7b04d8705495a2d01f4a11e51.

## Skill

- [Building with Oort (SKILL.md)](https://oortcloud.dev/llms/skills/oort-sdk/SKILL.md): the oort-sdk Agent Skill: which integration to use, the agent loop, the rules that are not optional
- [Hooks, MCP, SDK or CLI](https://oortcloud.dev/llms/skills/oort-sdk/references/integrations.md)
- [The TypeScript SDK: modes, options, methods](https://oortcloud.dev/llms/skills/oort-sdk/references/sdk.md): Modes (`local`, `cloud`, `auto`), every option and method, results, ids, limits, lifecycle
- [Multi-user and multi-tenant apps](https://oortcloud.dev/llms/skills/oort-sdk/references/multi-user.md): A multi-user or multi-tenant app: `scoped({ user })`, `org`, what is refused
- [Judgments: an expert's decisions, applied to new items](https://oortcloud.dev/llms/skills/oort-sdk/references/judgments.md): Expert judgments: `judge.record`, `judge.apply`, `judge.confirm`, agent and expert roles
- [Errors](https://oortcloud.dev/llms/skills/oort-sdk/references/errors.md): Errors: classes, `code`, `reason`, `fix`, `retryable`, `retryAfter`, a retry helper
- [Frameworks](https://oortcloud.dev/llms/skills/oort-sdk/references/frameworks.md): Vercel AI SDK (`oortmem/ai-sdk`), LangGraph.js, OpenAI Agents JS; Python LangGraph, OpenAI Agents, CrewAI
- [Services, serverless and self-hosting](https://oortcloud.dev/llms/skills/oort-sdk/references/deployment.md): A long-running service, serverless (Lambda, Cloud Functions, Cloud Run), bundling
- [Which calls cost usage](https://oortcloud.dev/llms/skills/oort-sdk/references/costs.md): Which calls bill a model vendor, the Oort Pro relay or Oort Cloud
- [Python, and the CLI from scripts and other languages](https://oortcloud.dev/llms/skills/oort-sdk/references/cli-and-python.md): Python (the native client), shell scripts, CI: the CLI with `--json`, `oort.v1`, exit codes
- [Security](https://oortcloud.dev/llms/skills/oort-sdk/references/security.md): Keys, secret scrubbing, prompt injection, scopes and trust
- [Installing the skill](https://oortcloud.dev/llms/skills/README.md): Claude Code and Codex plugins, or copy the folder into any agent that reads SKILL.md

## Docs

- [Oort from code: the TypeScript SDK](https://oortcloud.dev/llms/docs/sdk.md): the TypeScript SDK guide: createOort, modes, the agent loop
- [Oort from Python: the native client](https://oortcloud.dev/llms/docs/python.md): the native Python client: from oortmem import Oort, sync and async
- [`oortmem` API reference](https://oortcloud.dev/llms/docs/sdk-reference.md): every name the oortmem package exports, generated from its declarations

## Optional

- [The `oort` command line](https://oortcloud.dev/llms/docs/cli.md): every oort command and flag, and the --json output (152 KB)
- [Tested examples](https://github.com/oortmem/oort/tree/v0.9.0/examples): the TypeScript and Python examples the skill and docs quote, run in CI