# Hope Agent — Open-source AI assistant with memory and tools Canonical page: https://www.hopeagent.ai/en An AI assistant that gets work done and remembers you. Remembers you, grows over time, pursues goals autonomously, orchestrates work dynamically, and stays reachable from every chat app you use. The native installer includes templates for major model providers—paste an API key or sign in and start chatting, with no runtime, CLI, or config wrangling. ## Completes work ### Goal · keep pursuing the outcome Give Hope Agent an outcome and completion criteria, and it keeps decomposing, executing, checking, and advancing—with budgets, pause/resume, and a conservative evidence-backed completion audit. ### Workflow · dynamic orchestration The model organizes phases, conditions, parallel work, multiple agents, tools, diffs, reviews, and verification as needed, with durable pause, resume, cancel, and conservative recovery. ### Loop · continue on time or events Continue on intervals, conditions, internal events, or model-selected wakeups. Each iteration can resume the chat or launch a Goal-bound Workflow, with budgets, backoff, and no-progress protection. ### Side chats & tabbed panels Ask a separate question with /side while the main task runs. Reference past chats with @ or import read-only local Codex history. Files, diffs, plans, and the browser share reorderable right-side tabs. ## Design, knowledge & growth ### Design Space Generate 10+ artifact types—including websites, mobile prototypes, and presentations—from a prompt, image, or URL, with live preview, visual editing, annotations, version history, and multi-format export. ### Design system to code Extract a brand system from screenshots, URLs, Figma, or a repository, reuse design tokens, and hand selected artifacts into a real codebase. ### Knowledge Space · second brain You and the AI work on real Markdown with full-text and vector retrieval, backlinks, graph view, atomic notes, and reviewable organization proposals; existing Obsidian vaults can be attached. ### Long-term memory & skills Global, Project, and Agent memory returns on demand and consolidates offline into a profile; complex work can become reusable agentskills.io-compatible skills. ## Tools, connections & handoff ### Computer & browser control With permission, operate the macOS desktop, windows, menus, keyboard, and pointer, while the controllable browser mirrors web actions live; side effects share one approval flow. ### Multiple agents & scheduling Preset teams or dynamic sub-agents collaborate. Scheduled work can create a new chat or continue an existing one, with run history, errors, and delivery results. ### MCP, Hooks & workspaces Connect services through MCP, customize behavior with lifecycle Hooks, use Feishu/Lark workspace tools, and preview skills from GitHub or local folders before installing. ### Cross-device & native streaming IM Desktop, browser, and popular IMs share sessions, memory, and task state; Slack and Telegram stream natively into one reply, while messages sent to busy sessions queue durably in order. ## Local-first & dependable ### Local-first · zero third-party hops Configuration, sessions, memory, attachments, skills, and logs stay under ~/.hope-agent/ by default; model requests go directly to providers, with Owner Token auth and SSRF protection in server mode. ### Tool approval + Docker sandbox Sensitive tool calls enter one approval flow; high-risk commands and writes can run inside a Docker sandbox, including isolated workspaces in container deployments. ### Rollback, recovery & keepalive Configuration changes create local snapshots for one-click rollback; Guardian, system services, and subsystem watchdogs provide restart, diagnosis, and reconnection. ### Dashboard, Recap & evaluation Track cost, tokens, health, and long-running work; generate conversation recaps and evaluate Goal, Workflow, async-task, and multi-agent scenarios with real models. ## Batteries included 50 Provider templates; 481 preset model entries. Telegram, Discord, Slack, Feishu, Google Chat, LINE, QQ Bot, Signal, iMessage, IRC, WeChat, WhatsApp Simplified Chinese, Traditional Chinese, English, Japanese, Korean, Spanish, Portuguese, Russian, Arabic, Turkish, Vietnamese, Malay ## Watch Hope Agent in 3 minutes English narration and captions. Explore goals, tools, memory, knowledge, and design in one desktop workflow. Product presentation built with real Hope Agent components and fixed demo data. [Watch Hope Agent in 3 minutes](https://github.com/user-attachments/assets/e1728465-0f5b-47a6-8af8-b9806067a6b4) 1920×1080 screenshots of the running Web GUI. Projects, sidebar entries, and notes use demo content; deliverables come from actual runs. Dashboard metrics use illustrative sample data. ### Chat & deliverables From a project brief to three deliverable drafts, with project history and session status in view. ### Design Space A generated landing page with an editorial hero, product illustration, feature sections, and a call to action. ### Knowledge Space Nine connected Markdown notes inform an AI creative brief, with the source note and response side by side. ### Dashboard A 30-day view of usage, cost, health, and activity, with comparisons to the previous period. Metrics use illustrative demo data. ## Documentation and source - [Install Hope Agent](https://www.hopeagent.ai/en/install) - [Hope Agent User Guide](https://www.hopeagent.ai/en/guide) - [Running modes](https://www.hopeagent.ai/en/running-modes) - [Features](https://www.hopeagent.ai/en/features) - [Self-Hosting](https://www.hopeagent.ai/en/self-host) - [Frequently asked questions](https://www.hopeagent.ai/en/faq) - [What's new](https://www.hopeagent.ai/en/changelog) - [GitHub](https://github.com/shiwenwen/hope-agent) - [Releases](https://github.com/shiwenwen/hope-agent/releases) - [MIT](https://github.com/shiwenwen/hope-agent/blob/main/LICENSE)