OpenWarp

A fully decentralized terminal — your AI, your agents, your keys, your machine.

OpenWarp is a community fork of Warp that strips Warp's mandatory cloud dependency while preserving the full Warp terminal experience. It opens up the AI provider layer, lets you plug in any third-party CLI agent, ships a built-in SSH host manager, and fixes a number of upstream rendering issues — all while keeping every credential, conversation, and agent history on your own machine.

简体中文 · Upstream Warp · Upstream sync notes

Early development. No official release yet. Not affiliated with Warp, Inc.

---

Why OpenWarp

Upstream Warp ties AI, accounts, sync, and agent history to Warp's cloud. OpenWarp opens that layer entirely and adds capabilities the upstream client does not provide:

	Upstream Warp	OpenWarp
Cloud dependency	Hard dependency on Warp backend (auth / Drive / history / Agent)	Fully decentralized, no mandatory cloud calls
AI provider	Warp gateway only	Any OpenAI-compatible endpoint + 6 native protocols
Third-party agents	Built-in Warp Agent only	Any CLI agent — DeepSeek-TUI / Codex / Claude Code wired in
SSH management	Not built-in	Built-in SSH host manager (connect / config / tmux)
Markdown rendering	Upstream baseline	Tuned MD pipeline — code blocks, tables, mixed CJK
Font rendering	Upstream cosmic_text default	CJK soft-wrap caret + bold subpixel fixes
Credentials	Cloud account	Local config file, never leaves device
System prompt	Server-assembled, opaque	minijinja templates, fully editable
UI language	English	Native English + Simplified Chinese, extensible
Cloud Agent / Computer Use	On by default	Off by default (and being physically removed)
Blocks / Workflows / Keymaps	Kept	Fully preserved, continuously synced
License	AGPL-3.0 / MIT dual	Same as upstream

Things upstream Warp does NOT support, but OpenWarp does

These are net-new capabilities OpenWarp adds on top of the fork:

• SSH host manager — connect, configure and manage SSH hosts and sessions directly inside the terminal (with tmux integration). No external switcher needed.
• Third-party CLI agents — bring any CLI agent into the Warp Block model. First-class adapters for:
  • DeepSeek-TUI (completion notifications, text-notification mapping, input-restore plumbing all wired up)
  • Codex CLI, Claude Code, and other mainstream CLI agents
  • Unified routing through OSC9 / OSC777 into Warp's notification center
• BYOP across many providers — 6 native protocols (OpenAI / OpenAIResp / Anthropic / Gemini / Ollama / DeepSeek) explicitly bound; any OpenAI-compatible proxy works out of the box. Credentials stay local.
• Fully decentralized — no Warp account, no forced login, no cloud Drive / Notebook sync, no cloud agent history. Cloud code paths are being physically removed in stages.
• Markdown rendering improvements — better stability for code blocks, tables, lists and mixed Chinese/English text inside AI blocks.
• Font rendering algorithm fixes — CJK soft-wrap caret offset, bold-on-small Chinese characters, and other long-standing upstream rendering papercuts.

Three steps to take your terminal fully into your own hands

01 · Plug in any provider Paste a Base URL and API key in settings — any OpenAI Chat Completions–compatible endpoint works out of the box. Credentials are stored locally only.

02 · Author dynamic prompts A minijinja-powered template engine renders the system prompt in real time based on the current working directory, language, and role.

03 · Use it immediately Switch models, conversations, command suggestions, and third-party agents with one click — the experience is identical to Warp, but every layer is yours.

Verified AI providers

Provider	Base URL	Notes
OpenAI	https://api.openai.com/v1	Native protocol
Anthropic	via genai native	Claude 4.x family
DeepSeek	https://api.deepseek.com/v1	thinking + tool calling
Gemini	via genai native	Google AI Studio
Ollama	http://localhost:11434/v1	Local inference, no key
OpenRouter	https://openrouter.ai/api/v1	Aggregator gateway
Qwen / Groq / Together / LM Studio / any OpenAI-compatible proxy	—	Configure and go

Core features

• BYOP custom providers — 6 native protocols explicitly bound on top of genai 0.6
• Third-party CLI agents — DeepSeek-TUI / Codex CLI / Claude Code routed through OSC9 into Blocks and the notification center
• SSH host manager — manage SSH hosts and sessions inside the terminal, with tmux integration
• SSE streaming — incremental block rendering identical to Warp's first-party path
• 18 local tools — shell / read / edit / search / mcp / drive docs / skills / ask, all executed locally
• System prompt templates — eight model-family prompts ported from opencode (default / anthropic / gpt / beast / gemini / kimi / codex / trinity)
• models.dev integration — searchable Providers subpage with thousands of preloaded model entries
• Rendering improvements — tuned Markdown pipeline + CJK soft-wrap / bold fixes
• Privacy first — Cloud Agent / Computer Use / Referral / telemetry all disabled by default
• Warp experience preserved — continuously merged with upstream; Blocks, Workflows, AI commands, Keymaps and themes all kept
• Localized UI — Simplified Chinese + English, community-extensible

What we are aiming for

OpenWarp wants to be the kind of terminal that:

1. Runs fully without any centralized service — no account, no forced login, no feature that "only works when the cloud is reachable".
2. Treats AI and agents as an open ecosystem, not a single vendor — every mainstream LLM provider and CLI agent is a first-class citizen.
3. Makes remote work native — SSH / tmux / remote sessions are built-in, not bolted on.
4. Earns the right to be used all day — mixed CJK, Markdown, code blocks and font rendering should never be the weak link.
5. Stays in sync with upstream Warp — benefit from Warp's engineering work while keeping fork-level autonomy on direction.

If you share these goals, come help us finish it.

Build from source

git clone https://github.com/zerx-lab/openwarp
cd openwarp
./script/bootstrap   # platform-specific deps
./script/run         # build & run
./script/presubmit   # fmt / clippy / tests

If you prefer raw cargo, always target the OSS binary explicitly:

cargo build --release --bin warp-oss
cargo run   --release --bin warp-oss

Do not run cargo build --release / cargo run --release --bin {warp,stable,dev,preview} without a filter — those entry points (local.rs / stable.rs / dev.rs / preview.rs) load their channel config through Warp's private warp-channel-config binary, which lives in a closed-source repo. Compilation succeeds, but the resulting executables panic at startup asking you to run ./script/install_channel_config. That script clones an SSH repo only Warp employees can access. OpenWarp users only need the warp-oss binary.

See WARP.md for the full engineering guide (style, testing, platform notes).

License

Same as upstream Warp:

• warpui_core / warpui crates — MIT
• Everything else — AGPL-3.0

Branches & upstream sync

zerx-lab/warp keeps two long-lived branches:

Branch	Tracks	Purpose
main	zerx-lab/warp:main (default)	OpenWarp's main development line. All PRs target this.
warp-upstream	warpdotdev/warp:master	Pristine mirror of upstream Warp, used to pull in new commits. No fork-local changes.

For contributors

Open PRs against main. Never against warp-upstream.

For maintainers (write access)

Do not click the "Sync fork" button on main in the GitHub web UI. It would merge the entire upstream history straight into OpenWarp's main line and trigger large-scale conflicts. Pull upstream changes through the mirror branch, following the blacklist and flow in docs/openwarp-upstream-sync.md:

# one-time setup
git remote add upstream https://github.com/warpdotdev/warp.git

# refresh the mirror
git checkout warp-upstream
git pull                          # fast-forwards from upstream/master
git push origin warp-upstream

# bring selected commits into main
git checkout main
git cherry-pick <sha>             # or merge warp-upstream when a full sync makes sense

Featured Partner

DeepSeek-TUI — a terminal UI for the DeepSeek model family. OpenWarp ships first-class integration: completion notifications, OSC9 text-notification mapping, and input-restore plumbing are all wired up so DeepSeek-TUI runs as a native Block inside OpenWarp.

Launch it with deepseek from any OpenWarp terminal — Block lifecycle, footer status and notification center all work out of the box.

Windows note — DeepSeek-TUI's [notifications].method defaults to auto, which resolves to Off on Windows for any TERM_PROGRAM outside its built-in allowlist (iTerm.app / Ghostty / WezTerm). OpenWarp identifies as WarpTerminal, so to receive turn-completion notifications inside OpenWarp on Windows, add the following to ~/.deepseek/config.toml:

[notifications]
method = "osc9"

[tui]
notification_condition = "always"  # optional: notify on every turn

If you maintain a CLI agent or a terminal-adjacent tool and want similar first-class integration, open an issue — we are happy to wire more partners in.

Contributing

Community contributions welcome. See CONTRIBUTING.md for the full flow.

Before filing, please search existing issues. Security vulnerabilities should be reported privately per CONTRIBUTING.md#reporting-security-issues.

Acknowledgements

OpenWarp stands on the shoulders of the Warp team and many open-source projects:

Warp · genai · opencode · models.dev · DeepSeek-TUI · Codex CLI · Tokio · NuShell · Alacritty · Hyper · minijinja · cosmic-text