# Palette > Palette makes it practical to work with agents on your own files today, while building the connected OS for AI-native teams. Palette Desktop lets teams work with Claude Code, Codex, Gemini, and more on folders they control. Palette OS is the wider system Palette is building to connect company context, shared skills, and agent work across teams, tools, and AI sessions. For the complete public documentation and Marketplace corpus, see [llms-full.txt](https://palette.team/llms-full.txt). For live tool access to the documentation and Marketplace corpus, connect an MCP client to [https://palette.team/api/mcp](https://palette.team/api/mcp). Call `list_capabilities` first to see what Palette can do, then `search_docs` and `get_doc`. ## Pages - [Homepage](https://palette.team/): Work with supported agents on your own files today, and see the wider Palette OS direction. - [Palette Desktop](https://palette.team/desktop): Use Claude Code, Codex, Gemini, and more on folders you control, with safe sessions and review before saving. - [Palette OS](https://palette.team/os): Preview the system Palette is building for shared company context, skills, handoffs, and connected agent work. - [Pricing](https://palette.team/pricing): One flat price across your whole organization, no matter how many users, with a free plan and your own Claude or Codex subscriptions. - [Book a founder call](https://palette.team/book-a-call): Share your team and current need, then hear from Palette about a working session to review AI adoption and choose a useful next step. - [Design Partner Program](https://palette.team/design-partner): Apply for early access to help shape Palette's development. - [Windows Waitlist](https://palette.team/windows-waitlist): Join the waitlist for Palette Desktop on Windows. - [Press Kit](https://palette.team/press): Latest Palette announcements and coverage, plus paste-ready company descriptions, facts, logos, photos, screenshots, and brand guidance. ## Marketplace - [Marketplace](https://palette.team/marketplace): Free, ready-made folder systems and workflows for Palette Desktop. Pick one, install it, and let your agent adapt it to your work. - [Marketplace full text](https://palette.team/marketplace/llms-full.txt): Complete public descriptions, structures, commands, and FAQs for every active item. - [Marketplace catalog JSON](https://palette.team/marketplace/catalog.json): Machine-readable metadata and artifact facts for every active item. - [Company OS](https://palette.team/marketplace/company-os): Keep company context across accounts, GTM, product, engineering, operations, and people in one shared folder. - [GTM OS](https://palette.team/marketplace/gtm-os): Keep brand, market, strategy, accounts, campaigns, playbooks, team context, and GTM memory in one folder. - [Product OS](https://palette.team/marketplace/product-os): Turn feedback, interviews, requests, and research into clearer product decisions and paste-ready specs. - [Project OS](https://palette.team/marketplace/project-os): Run a project from briefs, meetings, decisions, research, deliverables, team context, and a generated status dashboard. - [Accounts](https://palette.team/marketplace/accounts): Keep account briefs, meeting notes, health reads, next steps, and call preparation together in plain files. - [Content](https://palette.team/marketplace/content): Turn ideas into on-brand drafts, distribution plans, and a clear record of what you shipped. - [Competitor Analysis](https://palette.team/marketplace/competitor-analysis): Maintain a market landscape and sourced competitor profiles with repeatable research and update workflows. - [Presentations](https://palette.team/marketplace/presentations): Turn Markdown drafts into polished HTML slide decks with presenter notes, themes, and PDF export. - [Workspace Heal](https://palette.team/marketplace/workspace-heal): Find broken links, stale indexes, unfinished setup, and instruction drift before they make your workspace harder to trust. ## Blog - [Blog](https://palette.team/blog): Read candid notes from the Palette team about what we are building, what we are learning, and how people and AI agents can work from the same context. - [We raised a €3M pre-seed to build the OS for AI-native teams](https://palette.team/blog/announcing-our-pre-seed): Ugly Duckling Ventures led our round, with participation from Emblem, Acadian Ventures, and a group of angels. Here's why we raised it and what comes next. - [Why We're Building Palette](https://palette.team/blog/hello-world): A short introduction to Palette — a shared context layer for your team and your AI, built by a team of six in Copenhagen. - [Launching Palette Desktop](https://palette.team/blog/launching-palette-desktop): Palette Desktop is here — a desktop app for running Claude Code and Codex on shared team folders. For the parts of the company that aren't engineering. - [We raised a €3M pre-seed to build the OS for AI-native teams](https://palette.team/blog/announcing-our-pre-seed): Ugly Duckling Ventures led our round, with participation from Emblem, Acadian Ventures, and a group of angels. Here's why we raised it and what comes next. ## Articles - [Articles](https://palette.team/articles): Practical, experience-based articles on AI agents, shared context, and team workflows, drawn from systems the Palette team has built and used in real work. - [How we gave our agents a memory with lesson and log files](https://palette.team/articles/how-we-gave-our-agents-memory): How we gave our AI agents a memory that survives the session, using two plain markdown files: lessons.md for what to do differently, log.md for what happened. ## Changelog - [Changelog](https://palette.team/changelog): Latest updates, improvements, and fixes to Palette. - [Palette Desktop 0.28.0](https://palette.team/changelog/desktop-v0.28.0-cloud-and-self-hosted-models): Run the Palette agent on a Palette-hosted cloud model or your own self-hosted server. With on-device, that's three model sources, chosen per tab, plus external HTML share links. - [Palette Desktop 0.27.0](https://palette.team/changelog/desktop-v0.27.0-on-device-agent): Palette can now run a local model through a built-in on-device agent. Starting with Qwen, it works offline, and nothing you type ever leaves your device. - [Palette Desktop 0.26.0](https://palette.team/changelog/palette-desktop-0-26-0): Gemini joins Claude Code, Codex, and Mistral, one-click install for every provider, plus early HTML sharing and handoffs. - [Palette Desktop 0.21.0](https://palette.team/changelog/palette-desktop-0-21-0): First update since launch — a new Mistral Vibe provider, hands-on file management in the Files area, Palette account sign-in, and a stack of workspace polish. - [Palette Desktop 0.1.0](https://palette.team/changelog/palette-desktop): Palette Desktop launches — run Claude Code and Codex on shared team folders, with sessions, agent choice, templates, and Palette context built in. - [Briefs](https://palette.team/changelog/briefs): The morning brief your chief of staff would write — TLDR, what needs attention, and today's calendar, delivered to your team's Slack. - [Palette v1.0](https://palette.team/changelog/initial-launch): The first release of the Palette context layer — connect your tools, build a living context map of your org, and serve it to your AI tools via MCP. ## Jobs - [Jobs](https://palette.team/jobs): Open roles at Palette — help build the context layer for agentic organizations. - [Product Engineer (Apps)](https://palette.team/jobs/product-engineer-apps): Drive Palette Desktop end to end — the app teams use to put their context to work. The first hire dedicated to Desktop. An engineering role in Copenhagen. - [Founding Data/ML Engineer](https://palette.team/jobs/founding-data-ml-engineer): Own the signal engine — take Palette's context brain from events and signals to a queryable graph of everything that happens across a company's tools. A founding engineering role in Copenhagen. - [Founding Product Designer](https://palette.team/jobs/founding-product-designer): A founding, full-stack design role in Copenhagen — this position has been filled. - [Founding Customer Lead](https://palette.team/jobs/founding-customer-lead): Our first commercial hire — onboard, activate, and grow our earliest customers, then build the playbook and engine that scales it without us in the room. A founding GTM role in Copenhagen. ## Documentation — About Palette - [Welcome to Palette](https://palette.team/docs/welcome-to-palette): Meet Palette Desktop, available today for agent work on files, and Palette OS, the wider system Palette is building for shared context, skills, and agent work. - [Company](https://palette.team/docs/about-palette): Palette is based in Copenhagen, Denmark. Four co-founders building Palette Desktop and Palette OS, backed by Emblem, Acadian Ventures, and Ugly Duckling Ventures. - [Vision & Manifesto](https://palette.team/docs/palette-manifesto-why-context-is-infrastructure): Palette's vision and manifesto: why knowledge work is changing, why individuals now move faster than organizations, and why AI-native teams need shared context. - [FAQ](https://palette.team/docs/frequently-asked-questions): Answers about Palette Desktop, Palette OS, security, early access, and how the Context Library makes scoped company context available to people and agents. - [AI and agent glossary for teams](https://palette.team/docs/ai-agent-glossary-for-teams): Plain-English definitions of AI, agent, teamwork, and Palette terms, including tokens, context windows, MCP, workspaces, sessions, and shared context. ## Documentation — Palette Desktop - [Palette Desktop](https://palette.team/docs/palette-desktop-overview): Palette Desktop is a desktop app for working with Claude Code, Codex, Gemini, and other agents on your own files and folders. Use safe sessions and review every change before saving it back. - [Quick Start](https://palette.team/docs/palette-desktop-quickstart): Get started with Palette Desktop in minutes. Download the app, point it at a folder on Google Drive or Dropbox, run your first session, and review changes before saving back. - [Workspaces](https://palette.team/docs/palette-desktop-workspaces): A workspace in Palette Desktop is a folder on your computer. Point Palette at it, and the folder becomes a workspace your agents can read, write, and reason about. - [Sharing workspaces](https://palette.team/docs/palette-desktop-sharing-workspaces): Put your workspace folder on Google Drive, Dropbox, or iCloud. Teammates work in separate sessions, then review and save their changes back. - [Sessions](https://palette.team/docs/palette-desktop-sessions): A session in Palette Desktop is a sandboxed clone of your workspace. The agent works inside the clone, you review every change, then you save the session back to the main folder. - [Chats](https://palette.team/docs/palette-desktop-chats): Chats are conversations with an agent inside a session. Run several in parallel, pick a different agent or model per chat, and pull specific files into context with @-references. - [Plan mode](https://palette.team/docs/palette-desktop-plan-mode): Plan mode is available with Claude Code and Palette's own agent. Review and adjust the agent's approach before it starts changing files. - [Actions](https://palette.team/docs/palette-desktop-actions): The actions list surfaces workspace skills and workflows. Run one from a chat, with slash-command and automatic-discovery behavior depending on the agent. - [Shortcuts](https://palette.team/docs/palette-desktop-shortcuts): Pin files, actions, and custom prompts in Palette Desktop. Your most-used things in the workspace, one click away from any chat. - [References](https://palette.team/docs/palette-desktop-references): References attach folders outside the current workspace to a chat. Set each reference to read-only or editable, depending on what the agent should be allowed to do. - [Marketplace](https://palette.team/docs/palette-desktop-templates): Install a folder system or workflow from Marketplace into a current or new session, then follow the setup chat and use the checkpoint if you want to undo. - [Agents](https://palette.team/docs/palette-desktop-agents): Palette Desktop runs five agents: Palette's own agent plus Claude Code, Codex, Gemini CLI, and Mistral Vibe. Pick one per chat, and choose how models are powered. - [Palette agent](https://palette.team/docs/palette-desktop-palette-agent): Palette's own built-in agent. Pick a model and choose where it runs: Palette Cloud, an on-device model, or your own endpoint. - [Claude Code](https://palette.team/docs/palette-desktop-claude-code): Run Anthropic's Claude Code in Palette Desktop, with Opus, Sonnet, Haiku, and Fable models, signed in with your Claude subscription or with your own Anthropic API key. - [Codex](https://palette.team/docs/palette-desktop-codex): Run OpenAI's Codex in Palette Desktop. Sign in with your ChatGPT or Codex plan, or add an OpenAI API key. The models you can pick come from your own plan, not a fixed Palette list. - [Gemini](https://palette.team/docs/palette-desktop-gemini): Run Google's Gemini in Palette Desktop. Gemini uses a Google AI Studio API key (not a Google sign-in). Models: Auto, Gemini Pro, Gemini Flash. - [Mistral Vibe](https://palette.team/docs/palette-desktop-mistral): Run Mistral's Vibe agent in Palette Desktop. Models: Mistral Medium and Devstral Small. Signing in with Mistral provisions a La Plateforme API key, billed to your own Mistral account. - [CLAUDE.md and AGENTS.md](https://palette.team/docs/palette-desktop-working-with-claude-md): CLAUDE.md gives Claude Code workspace instructions, while AGENTS.md does the same for Codex. Other agents may need you to mention the file explicitly. - [Working with README.md files](https://palette.team/docs/palette-desktop-readme-and-index-files): A short README or index file tells your team what a folder is for and gives agents useful context when you link or point them to it. - [Structuring your workspace](https://palette.team/docs/palette-desktop-structuring-your-workspace): A folder is how you give an agent context. A clear, shallow structure named for the work helps agents and teammates find things fast. - [Working with lessons.md](https://palette.team/docs/palette-desktop-working-with-lessons): lessons.md is a shared list of durable corrections. Link it from your agent instructions or ask the agent to read it when you want those lessons applied. - [Working with log.md](https://palette.team/docs/palette-desktop-working-with-logs): log.md is a short, shared record of the decisions and changes that matter. It is team memory kept in the workspace, focused enough to stay useful instead of bloating context. - [The viewer](https://palette.team/docs/palette-desktop-the-viewer): The viewer is where you see and work with your files inside Palette Desktop: read and edit markdown, preview HTML, and view your other files, right next to the chat that produced them. - [Editing markdown in the viewer](https://palette.team/docs/palette-desktop-editing-markdown): Read and edit your markdown files right in Palette Desktop. The viewer shows them as clean, formatted documents you can type into, and they stay plain markdown on disk. - [Working with HTML files](https://palette.team/docs/palette-desktop-working-with-html-files): HTML is one of the best output formats for AI-generated work. Ask the agent for HTML, preview it inside Palette Desktop, and share it as a real artifact. - [FAQ](https://palette.team/docs/palette-desktop-faq): Common questions about Palette Desktop: which agents it runs, whether you need a subscription, where your data lives and how it stays private, review and undo, team collaboration, and how it compares to Cowork, Cursor, and chat tools. - [Context Library in Palette Desktop](https://palette.team/docs/palette-desktop-context-layer): In Palette Desktop, browse Context Library pages and mention one in chat. The agent can read the selected page when the Palette connector is connected. ## Documentation — Palette OS - [Overview](https://palette.team/docs/welcome-to-palette-shared-context-for-teams-and-ai): The Context Library is the Palette OS feature that turns selected company activity and check-ins into scoped context for people and agents. - [Quick start](https://palette.team/docs/quick-start-guide-set-up-palette): Set up the Context Library by connecting supported company tools, generating scoped context, and making it available in Palette Desktop or through MCP. - [How the Context Library works](https://palette.team/docs/how-palette-works-connected-tools-to-living-context): The Context Library captures selected activity and check-ins, refreshes scoped context each week, and makes it available through Palette Desktop and MCP. - [Context map](https://palette.team/docs/context-living-map-of-your-organization): The Context Library organizes generated company context across organization, team, and user scopes, with pages refreshed weekly for people and agents. - [Signal engine](https://palette.team/docs/signal-engine-how-palette-turns-events-into-context): Palette captures selected company activity throughout the week, combines it with check-ins during weekly generation, and refreshes scoped Context Library pages. - [Use Context Library with AI tools](https://palette.team/docs/palette-mcp-live-context-for-ai-tools): Use Palette MCP to let supported agents search and read permitted Context Library pages on demand, with practical prompts for company and team context. - [Use Context Library in workflows](https://palette.team/docs/optimise-your-workflows-with-palette): Use permitted Context Library pages in external agent workflows through Palette MCP, with patterns for digests, meeting preparation, onboarding, and planning. - [MCP](https://palette.team/docs/mcp-connect-palette-to-claude-chatgpt-ai-tools): Palette MCP gives supported agents on-demand access to the Palette OS context and capabilities permitted for the signed-in person. - [Connect to Claude](https://palette.team/docs/connect-palette-to-claude): Connect Claude and Claude Code to Palette through MCP so they can search and read the Palette OS context permitted for you on demand. - [Connect to ChatGPT](https://palette.team/docs/connect-palette-to-chatgpt): Connect ChatGPT to Palette OS using a custom MCP app when your plan and workspace allow it, with a manual Project snapshot as a fallback. - [Connect other tools](https://palette.team/docs/connect-palette-to-cursor-conductor-mcp-tools): Connect tools that support remote HTTP MCP and OAuth to Palette OS so agents can search and read the material permitted for you on demand. - [Slack](https://palette.team/docs/slack-integration): Palette connects to your Slack workspace and captures text messages, channel metadata, and user profiles from the channels you choose. - [Jira](https://palette.team/docs/jira-integration): Palette connects to Jira Cloud and captures issues, comments, changelogs, and user activity from the projects you choose. - [Linear](https://palette.team/docs/linear-integration): Palette connects to Linear and captures issues, comments, projects, cycles, teams, users, and initiatives when initiative access is enabled. - [Notion](https://palette.team/docs/notion-integration): Palette connects to Notion and captures page content, database metadata, comments, and users from the spaces you choose. - [GitHub](https://palette.team/docs/github-integration): Palette connects to GitHub, reads PRs, issues, reviews, and deployments from repos you choose. It reads activity only, never source code or file contents. - [Google Calendar](https://palette.team/docs/google-calendar-integration): Google Calendar is a personal Connector Gateway connection that lets supported agents read and manage calendars under the signed-in user's Google access. - [Context Library FAQ](https://palette.team/docs/palette-context-layer-faq): Common questions about the Context Library, including its weekly refresh, scoped access, supported company connections, and how people and agents read context. ## Legal - [Terms of Service](https://palette.team/terms): Terms governing use of the Palette website and services. - [Privacy Policy](https://palette.team/privacy): How Palette collects, uses, and protects personal information. - [Cookie Policy](https://palette.team/cookies): How Palette uses cookies and how to control them. - [Data Processing Agreement](https://palette.team/dpa): How Palette processes personal data on your behalf under the GDPR, including sub-processors, security measures, and international transfers.