CodeWhale

The most agentic harness for DeepSeek V4. Rules, tools, evidence, and feedback loops that help the model keep working until the task is done — and keep getting better at it.

DeepWiki project index

简体中文 README 日本語 README

Install · Quickstart · Usage · Documentation · Contributing · Support

Install

codewhale is distributed as Rust binaries: the dispatcher command (codewhale) and the companion TUI runtime (codewhale-tui). Pick whichever install path you already use; they all put the same commands on your PATH. The npm package is an installer/wrapper for the release binaries, not the agent runtime itself.

# 1. npm — easiest if you already use Node. The package downloads the
#    matching prebuilt Rust binaries from GitHub Releases.
npm install -g codewhale

# 2. Cargo — no Node needed. Requires Rust 1.88+ (the crates use the
#    2024 edition; older toolchains fail with "feature `edition2024` is
#    required"). Run `rustup update` first, or use a non-Cargo path below.
cargo install codewhale-cli --locked   # `codewhale` (entry point)
cargo install codewhale-tui     --locked   # `codewhale-tui` (TUI binary)

# 3. Homebrew — macOS package manager.
brew tap Hmbown/deepseek-tui
brew install deepseek-tui

# 4. Direct download — no package manager or toolchain.
#    https://github.com/Hmbown/CodeWhale/releases
#    Prebuilt for Linux x64/ARM64, macOS x64/ARM64, Windows x64.

# 5. Docker — prebuilt release image.
docker volume create codewhale-home
docker run --rm -it \
  -e DEEPSEEK_API_KEY="$DEEPSEEK_API_KEY" \
  -v codewhale-home:/home/codewhale/.deepseek \
  -v "$PWD:/workspace" \
  -w /workspace \
  ghcr.io/hmbown/codewhale:latest

In mainland China, speed up the npm path with --registry=https://registry.npmmirror.com, or use the Cargo mirror below.

Download safety: official release binaries live under https://github.com/Hmbown/CodeWhale/releases. For manual downloads, verify the SHA-256 manifest and avoid look-alike repositories or search-result mirrors. See download safety and checksums.

Already installed? Use the updater that matches the install path:

codewhale update                         # release-binary updater
npm install -g codewhale@latest      # npm wrapper
brew update && brew upgrade deepseek-tui
cargo install codewhale-cli --locked --force
cargo install codewhale-tui     --locked --force

---

What Is It?

A model answers a question. An agent finishes a task. The difference is the harness — the operating environment that surrounds the model with rules, tools, evidence, and feedback loops.

CodeWhale is that harness, built around DeepSeek V4 Pro and Flash. It started as a personal tool because the maintainer got tired of models losing track mid-task, obeying stale instructions over the user's current request, or giving up when a command failed. What emerged was a system that keeps the model oriented: a constitutional prompt hierarchy, structured trust boundaries, parallel sub-agents, prefix-cache-aware context management, and verification beats that give the model enough signal to self-correct.

DeepSeek V4 helped write parts of this harness. That matters because it means CodeWhale is already the most effective way to use V4 — and as V4 improves, the harness improves with it. Each turn leaves behind better prompts, better rules, and better handoffs. The next turn starts from a stronger position.

How the harness works

Agentic models deal with conflicting information at scale: user intent, project rules, system defaults, tool output, and stale memory all compete for authority in a single turn. LLM-as-a-judge needs jurisdiction — which source wins when they disagree?

CodeWhale answers this with a Constitution (prompts/base.md). It's a formal hierarchy of law — Article VII ranks nine sources from the Constitution's own articles down to prior-session handoffs. The user's current message outranks stale project instructions. Live tool output outranks assumptions. Verification outranks confidence. The model inherits a clear chain of authority every turn and never has to guess which directive to follow.

Seven articles sit above the hierarchy, defining the model's identity, duties, and agency: a verification mandate (Article V — every action leaves evidence, never declare success on faith), a coordination legacy (Article VI — leave the workspace legible for the next intelligence), and a primacy-of-truth clause (Article II — no lower rule may override it).

DeepSeek V4's prefix caching makes this practical. The Constitution is long and detailed, but once cached it costs roughly 100× less per turn than a cold read. The model references it recursively — peeking, scanning, and querying through RLM sessions — revisiting information on demand rather than relying on a single memorized pass. It performs more like an open-book test than a closed one.

Because the authority structure is explicit, failure isn't hidden. Non-zero exit codes, type errors from rust-analyzer arriving between turns, sandbox denials — these are fed back as correction vectors. The model uses its own drift to self-correct.

Three modes control the action space. Plan is read-only. Agent gates destructive operations behind approval. YOLO auto-approves in trusted workspaces. macOS Seatbelt is the active sandbox; Linux Landlock is detected but not yet enforced; Windows sandboxing is not yet advertised.

Fin — a cheap Flash call with thinking off — handles model auto-routing per turn. --model auto is the default.

Every turn records a side-git snapshot outside your repo's .git. /restore and revert_turn roll back the workspace.

Sub-agents run concurrently (up to 20). agent_open returns immediately; results arrive inline as completion sentinels with a summary. Full transcripts stay behind bounded handles through agent_eval. See docs/SUBAGENTS.md.

The rest of the surface: LSP diagnostics after every edit (rust-analyzer, pyright, typescript-language-server, gopls, clangd), RLM sessions for batched analysis, MCP protocol, HTTP/SSE runtime API, persistent task queue, ACP adapter for Zed, SWE-bench export, and live cost tracking with cache hit/miss breakdowns.

---

The Harness

codewhale (dispatcher CLI) → codewhale-tui (companion binary) → ratatui interface ↔ async engine ↔ OpenAI-compatible streaming client. Tool calls route through a typed registry (shell, file ops, git, web, sub-agents, MCP, RLM) and results stream back into the transcript. The engine manages session state, turn tracking, the durable task queue, and an LSP subsystem that feeds post-edit diagnostics into the model's context before the next reasoning step.

See docs/ARCHITECTURE.md for the full walkthrough.

Sub-agents: Concurrent Background Execution

CodeWhale can dispatch multiple sub-agents that run in parallel — like a concurrent task queue:

• Non-blocking launch. agent_open returns immediately. The child gets its own fresh context and tool registry and runs independently. The parent keeps working.
• Background execution. Sub-agents execute concurrently (default cap: 10, configurable to 20). The engine manages the pool — no polling loop needed.
• Completion notification. When a sub-agent finishes, the runtime injects a <codewhale:subagent.done> sentinel into the parent's transcript. The human-readable summary — including the child's findings, changed files, and any risks — sits on the line immediately before the sentinel. The parent model reads that summary and integrates findings without an extra tool call.
• Bounded result retrieval. The full child transcript lives behind a transcript_handle accessible through agent_eval. When the summary isn't enough, the parent calls handle_read for slices, line ranges, or JSONPath projections — keeping the parent context lean without losing access to the details.

See docs/SUBAGENTS.md for the full sub-agent reference.

---

Quickstart

npm install -g codewhale
codewhale --version
codewhale --model auto

Prebuilt binaries are published for Linux x64, Linux ARM64 (v0.8.8+), macOS x64, macOS ARM64, and Windows x64. For other targets (musl, riscv64, FreeBSD, etc.), see Install from source or docs/INSTALL.md.

On first launch you'll be prompted for your DeepSeek API key. The key is saved to ~/.deepseek/config.toml so it works from any directory without OS credential prompts.

You can also set it ahead of time:

codewhale auth set --provider deepseek   # saves to ~/.deepseek/config.toml
codewhale auth status                    # shows the active credential source

export DEEPSEEK_API_KEY="YOUR_KEY"      # env var alternative; use ~/.zshenv for non-interactive shells
codewhale

codewhale doctor                         # verify setup

If codewhale doctor says the rejected key came from DEEPSEEK_API_KEY, remove the stale export from your shell startup file, open a fresh shell, or run codewhale auth set --provider deepseek. Use codewhale auth status to see the config, keyring, and env-var source state without printing the key. Saved config keys take precedence over the keyring and environment and are easier to rotate.

To rotate or remove a saved key: codewhale auth clear --provider deepseek.

Tencent Cloud / CNB Remote-First Path

For an always-on workspace you can control from a phone, use the Tencent-native path: CNB mirror/source, Tencent Lighthouse HK, a Feishu/Lark long-connection bridge, and optional EdgeOne for a deliberate public HTTPS edge. The runtime API stays bound to localhost; EdgeOne is not used to expose /v1/*.

Start with docs/TENCENT_CLOUD_REMOTE_FIRST.md, then use docs/TENCENT_LIGHTHOUSE_HK.md for the server runbook.

Model Auto-Routing and Fin

Use codewhale --model auto or /model auto when you want codewhale to decide how much model and reasoning power a turn needs.

Model auto-routing controls two settings together:

• Model: deepseek-v4-flash or deepseek-v4-pro
• Thinking: off, high, or max

Before the real turn is sent, the app makes a small deepseek-v4-flash routing call with thinking off. That fast path is called Fin: a low-latency seam for model selection, summaries, RLM children, context maintenance, and other coordination work that should not spend a full reasoning turn. Fin looks at the latest request and recent context, then selects a concrete model and thinking level for the real request. Short/simple turns can stay on Flash with thinking off; coding, debugging, release work, architecture, security review, or ambiguous multi-step tasks can move up to Pro and/or higher thinking.

--model auto and /model auto are local to codewhale. The upstream API never receives model: "auto"; it receives the concrete model and thinking setting chosen for that turn. The TUI shows the selected route, and cost tracking is charged against the model that actually ran. If the Fin route fails or returns an invalid answer, the app falls back to a local heuristic. Sub-agents inherit model auto-routing unless you assign them an explicit model.

Use a fixed model or fixed thinking level when you want repeatable benchmarking, a strict cost ceiling, or a specific provider/model mapping.

Linux ARM64 (Raspberry Pi, Asahi, Graviton, HarmonyOS PC)

npm i -g codewhale works on glibc-based ARM64 Linux from v0.8.8 onward. You can also download prebuilt binaries from the Releases page and place them side by side on your PATH.

China / Mirror-friendly Installation

If GitHub or npm downloads are slow from mainland China, use a Cargo registry mirror:

# ~/.cargo/config.toml
[source.crates-io]
replace-with = "tuna"

[source.tuna]
registry = "sparse+https://mirrors.tuna.tsinghua.edu.cn/crates.io-index/"

Then install both binaries (the dispatcher delegates to the TUI at runtime):

cargo install codewhale-cli --locked   # provides `codewhale`
cargo install codewhale-tui     --locked   # provides `codewhale-tui`
codewhale --version

Prebuilt binaries can also be downloaded from GitHub Releases. Use DEEPSEEK_TUI_RELEASE_BASE_URL for mirrored release assets.

Windows (Scoop)

Scoop is a Windows package manager. The codewhale package is listed in Scoop's main bucket, but that manifest updates independently and can lag the GitHub/npm/Cargo release. Run scoop update first, then verify the installed version with codewhale --version:

scoop update
scoop install deepseek-tui
codewhale --version

Use npm or direct GitHub release downloads when you need the newest release before Scoop's manifest catches up.

Install from source

Works on any Tier-1 Rust target — including musl, riscv64, FreeBSD, and older ARM64 distros.

# Linux build deps (Debian/Ubuntu/RHEL):
#   sudo apt-get install -y build-essential pkg-config libdbus-1-dev
#   sudo dnf install -y gcc make pkgconf-pkg-config dbus-devel

git clone https://github.com/Hmbown/CodeWhale.git
cd CodeWhale

cargo install --path crates/cli --locked   # requires Rust 1.88+; provides `codewhale`
cargo install --path crates/tui --locked   # provides `codewhale-tui`

Both binaries are required. Cross-compilation and platform-specific notes: docs/INSTALL.md.

Other API Providers

Official DeepSeek remains the default and first-class path. Other providers are additive, with OpenRouter starting from DeepSeek Pro/Flash before broader open-model catalogs are enabled.

# NVIDIA NIM
codewhale auth set --provider nvidia-nim --api-key "YOUR_NVIDIA_API_KEY"
codewhale --provider nvidia-nim

# AtlasCloud
codewhale auth set --provider atlascloud --api-key "YOUR_ATLASCLOUD_API_KEY"
codewhale --provider atlascloud

# Wanjie Ark
codewhale auth set --provider wanjie-ark --api-key "YOUR_WANJIE_API_KEY"
codewhale --provider wanjie-ark --model deepseek-reasoner

# OpenRouter
codewhale auth set --provider openrouter --api-key "YOUR_OPENROUTER_API_KEY"
codewhale --provider openrouter --model deepseek/deepseek-v4-pro

# Novita
codewhale auth set --provider novita --api-key "YOUR_NOVITA_API_KEY"
codewhale --provider novita --model deepseek/deepseek-v4-pro

# Fireworks
codewhale auth set --provider fireworks --api-key "YOUR_FIREWORKS_API_KEY"
codewhale --provider fireworks --model deepseek-v4-pro

# Moonshot/Kimi
codewhale auth set --provider moonshot --api-key "YOUR_MOONSHOT_OR_KIMI_API_KEY"
codewhale --provider moonshot --model kimi-k2.6

# Generic OpenAI-compatible endpoint
codewhale auth set --provider openai --api-key "YOUR_OPENAI_COMPATIBLE_API_KEY"
OPENAI_BASE_URL="https://openai-compatible.example/v4" codewhale --provider openai --model deepseek-v4-pro

# Self-hosted SGLang
SGLANG_BASE_URL="http://localhost:30000/v1" codewhale --provider sglang --model deepseek-v4-flash

# Self-hosted vLLM
VLLM_BASE_URL="http://localhost:8000/v1" codewhale --provider vllm --model deepseek-v4-flash

# Self-hosted Ollama
ollama pull codewhale-coder:1.3b
codewhale --provider ollama --model codewhale-coder:1.3b

Inside the TUI, /provider opens the provider picker and /model opens the local model/thinking picker. /provider openrouter and /model <id> switch directly, while /models explicitly fetches and lists live API models when the active provider supports model listing.

---

Release Notes

Release-specific changes live in CHANGELOG.md. This README stays focused on current install paths, core workflows, provider setup, runtime interfaces, and extension points.

---

Usage

All examples use codewhale. The short form codew works everywhere — it's a silent alias that forwards to codewhale.

codewhale                                         # interactive TUI
codewhale "explain this function"                 # one-shot prompt
codewhale exec --auto --output-format stream-json "fix this bug"  # agentic exec with tool auto-approvals
codewhale swebench run --instance-id <ID> --issue-file issue.md  # write all_preds.jsonl for SWE-bench
codewhale exec --resume <SESSION_ID> "follow up"  # continue a non-interactive session
codewhale --model deepseek-v4-flash "summarize"   # model override
codewhale --model auto "fix this bug"             # auto-route model + thinking
codewhale --yolo                                  # auto-approve tools
codewhale auth set --provider deepseek            # save API key
codewhale doctor                                  # check setup & connectivity
codewhale doctor --json                           # machine-readable diagnostics
codewhale setup --status                          # read-only setup status
codewhale setup --tools --plugins                 # scaffold tool/plugin dirs
codewhale models                                  # list live API models
codewhale sessions                                # list saved sessions
codewhale resume --last                           # resume the most recent session in this workspace
codewhale resume <SESSION_ID>                     # resume a specific session by UUID
codewhale fork <SESSION_ID>                       # fork a saved session into a sibling path
codewhale serve --http                            # HTTP/SSE API server
codewhale serve --acp                             # ACP stdio adapter for Zed/custom agents
codewhale run pr <N>                              # fetch PR and pre-seed review prompt
codewhale mcp list                                # list configured MCP servers
codewhale mcp validate                            # validate MCP config/connectivity
codewhale mcp-server                              # run dispatcher MCP stdio server
codewhale update                                  # check for and apply binary updates

Branching Conversations

Saved sessions are intentionally branchable. codewhale fork <SESSION_ID> copies an existing saved session into a new sibling session, records the parent session id in metadata, and opens that fork so you can explore an alternate direction without polluting the original path. The session picker and codewhale sessions mark forked sessions with their parent id.

Inside the TUI, Esc-Esc backtrack can rewind the active transcript to a prior user prompt and put that prompt back in the composer for editing. /restore and revert_turn are separate workspace rollback tools: they restore files from side-git snapshots but do not rewrite conversation history.

Docker images are published to GHCR for release builds:

docker volume create codewhale-home

docker run --rm -it \
  -e DEEPSEEK_API_KEY="$DEEPSEEK_API_KEY" \
  -v codewhale-home:/home/codewhale/.deepseek \
  -v "$PWD:/workspace" \
  -w /workspace \
  ghcr.io/hmbown/codewhale:latest

See docs/DOCKER.md for pinned tags, local image builds, volume ownership notes, and non-interactive pipeline usage.

SWE-bench

CodeWhale can emit SWE-bench-compatible prediction JSONL from a checked-out task workspace:

codewhale swebench run \
  --instance-id django__django-12345 \
  --issue-file issue.md \
  --predictions-path all_preds.jsonl

run uses the same tool-backed automation path as codewhale exec --auto, then exports the final working-tree diff as model_patch. Use codewhale swebench export --instance-id <ID> when you have already produced the diff yourself. See docs/SWEBENCH.md for the full flow.

Zed / ACP

DeepSeek can run as a custom Agent Client Protocol server for editors that spawn local ACP agents over stdio. In Zed, add a custom agent server:

{
  "agent_servers": {
    "DeepSeek": {
      "type": "custom",
      "command": "codewhale",
      "args": ["serve", "--acp"],
      "env": {}
    }
  }
}

The first ACP slice supports new sessions and prompt responses through your existing DeepSeek config/API key. Tool-backed editing and checkpoint replay are not exposed through ACP yet.

Community-maintained adapter: acp-codewhale-adapter bridges codewhale exec --auto to cc-connect for users who need tool-backed ACP workflows outside the built-in Zed slice.

Keyboard Shortcuts

Key	Action
Tab	Complete / or @ entries; while running, queue draft as follow-up; otherwise cycle mode
Shift+Tab	Cycle reasoning-effort: off → high → max
F1	Searchable help overlay
Esc	Back / dismiss
Ctrl+K	Command palette
Ctrl+R	Resume an earlier session
Alt+R	Search prompt history and recover cleared drafts
Ctrl+S	Stash current draft (/stash list, /stash pop to recover)
@path	Attach file/directory context in composer
↑ (at composer start)	Select attachment row for removal

Full shortcut catalog: docs/KEYBINDINGS.md.

---

Modes

Mode	Behavior
Plan 🔍	Read-only investigation — model explores and proposes a plan before making changes; multi-step investigations use checklist_write
Agent 🤖	Default interactive mode — multi-step tool use with approval gates; substantial work is tracked with checklist_write
YOLO ⚡	Auto-approve all tools in a trusted workspace; multi-step work still keeps a visible checklist

Modes are separate from model auto-routing. Tab cycles Plan / Agent / YOLO, while /model auto controls model and thinking selection. The /goal command tracks a session objective and token budget today; a fuller Goal work surface is the right future home for persistent objective progress rather than another meaning of "auto".

---

Configuration

User config: ~/.deepseek/config.toml. Project overlay: <workspace>/.deepseek/config.toml (denied: api_key, base_url, provider, mcp_config_path). config.example.toml has every option.

Key environment variables:

Variable	Purpose
DEEPSEEK_API_KEY	API key
DEEPSEEK_BASE_URL	API base URL
DEEPSEEK_HTTP_HEADERS	Optional custom model request headers, e.g. X-Model-Provider-Id=your-model-provider
DEEPSEEK_MODEL	Default model
DEEPSEEK_STREAM_IDLE_TIMEOUT_SECS	Stream idle timeout in seconds, default 300, clamped to 1..=3600
DEEPSEEK_PROVIDER	codewhale (default), nvidia-nim, openai, atlascloud, wanjie-ark, openrouter, novita, fireworks, moonshot, sglang, vllm, ollama
DEEPSEEK_PROFILE	Config profile name
DEEPSEEK_MEMORY	Set to on to enable user memory
DEEPSEEK_ALLOW_INSECURE_HTTP=1	Allow non-local http:// API base URLs on trusted networks
NVIDIA_API_KEY / OPENAI_API_KEY / ATLASCLOUD_API_KEY / WANJIE_ARK_API_KEY / OPENROUTER_API_KEY / NOVITA_API_KEY / FIREWORKS_API_KEY / MOONSHOT_API_KEY / KIMI_API_KEY / SGLANG_API_KEY / VLLM_API_KEY / OLLAMA_API_KEY	Provider auth
OPENAI_BASE_URL / OPENAI_MODEL	Generic OpenAI-compatible endpoint and model ID
ATLASCLOUD_BASE_URL / ATLASCLOUD_MODEL	AtlasCloud endpoint and model override
WANJIE_ARK_BASE_URL / WANJIE_ARK_MODEL	Wanjie Ark endpoint and model override
MOONSHOT_BASE_URL / KIMI_BASE_URL / MOONSHOT_MODEL / KIMI_MODEL	Moonshot/Kimi endpoint and model override
OPENROUTER_BASE_URL	OpenRouter endpoint override
NOVITA_BASE_URL	Novita endpoint override
FIREWORKS_BASE_URL	Fireworks endpoint override
SGLANG_BASE_URL	Self-hosted SGLang endpoint
SGLANG_MODEL	Self-hosted SGLang model ID
VLLM_BASE_URL	Self-hosted vLLM endpoint
VLLM_MODEL	Self-hosted vLLM model ID
OLLAMA_BASE_URL	Self-hosted Ollama endpoint
OLLAMA_MODEL	Self-hosted Ollama model tag
NO_ANIMATIONS=1	Force accessibility mode at startup
SSL_CERT_FILE	Custom CA bundle for corporate proxies

Set locale in settings.toml, use /config locale zh-Hans, or rely on LC_ALL/LANG to choose UI chrome and the fallback language sent to V4 models. The latest user message still wins for natural-language reasoning and replies, so Chinese user turns stay Chinese even on an English system locale. See docs/CONFIGURATION.md and docs/MCP.md.

---

Models & Pricing

Model	Context	Input (cache hit)	Input (cache miss)	Output
deepseek-v4-pro	1M	$0.003625 / 1M	$0.435 / 1M	$0.87 / 1M
deepseek-v4-flash	1M	$0.0028 / 1M	$0.14 / 1M	$0.28 / 1M

DeepSeek Platform defaults to https://api.deepseek.com/beta so beta-gated API features can be tested without extra setup. Set base_url = "https://api.deepseek.com" to opt out.

Legacy aliases deepseek-chat / deepseek-reasoner map to deepseek-v4-flash and retire after July 24, 2026. NVIDIA NIM variants use your NVIDIA account terms.

Note

DeepSeek's pricing page now lists the V4 Pro rates above as the permanent prices: the previous 75% promotional discount has been folded into a one-quarter base-rate adjustment as the promotion window closes on 15:59 UTC on 31 May 2026. The TUI cost estimator already uses these values, so no behavioural change is required. For any future price changes, consult the official DeepSeek pricing page.

---

Publishing Your Own Skill

codewhale discovers skills from workspace directories (.agents/skills → skills → .opencode/skills → .claude/skills → .cursor/skills) and global directories (~/.agents/skills → ~/.claude/skills → ~/.deepseek/skills). Each skill is a directory with a SKILL.md file:

~/.agents/skills/my-skill/
└── SKILL.md

Frontmatter required:

---
name: my-skill
description: Use this when DeepSeek should follow my custom workflow.
---

# My Skill
Instructions for the agent go here.

Commands: /skills (list), /skill <name> (activate), /skill new (scaffold), /skill install github:<owner>/<repo> (community), /skill update / uninstall / trust. Community installs from GitHub require no backend service. Installed skills appear in the model-visible session context; the agent can auto-select relevant skills via the load_skill tool when your task matches their descriptions.

First launch also installs bundled system skills for common workflows: skill-creator, delegate, v4-best-practices, plugin-creator, skill-installer, mcp-builder, documents, presentations, spreadsheets, pdf, and feishu. These live under ~/.deepseek/skills and are versioned so new bundles are added on upgrade without recreating skills the user deliberately deleted.

---

Documentation

Doc	Topic
ARCHITECTURE.md	Codebase internals
CONFIGURATION.md	Full config reference
MODES.md	Plan / Agent / YOLO modes
MCP.md	Model Context Protocol integration
RUNTIME_API.md	HTTP/SSE API server
INSTALL.md	Platform-specific install guide
DOCKER.md	GHCR image, volumes, and Docker usage
CNB_MIRROR.md	CNB mirror and China-friendly install notes
TENCENT_CLOUD_REMOTE_FIRST.md	Tencent/CNB/Lighthouse/Feishu remote-first path
TENCENT_LIGHTHOUSE_HK.md	Lighthouse Hong Kong server setup
MEMORY.md	User memory feature guide
SUBAGENTS.md	Sub-agent role taxonomy and lifecycle
KEYBINDINGS.md	Full shortcut catalog
RELEASE_RUNBOOK.md	Release process
LOCALIZATION.md	UI locale matrix & switching
OPERATIONS_RUNBOOK.md	Ops & recovery
RECURSIVE_SELF_IMPROVEMENT.md	Copyable prompts for agent-assisted CodeWhale improvements

Full Changelog: CHANGELOG.md.

---

Support

CodeWhale is MIT-licensed and usable without sponsorship. If it saves you time, the clearest way to support ongoing maintenance is GitHub Sponsors. One-time support is also available through Buy Me a Coffee.

Sponsorship helps cover release builds, CI/runtime testing, package publishing, and maintainer time for issue triage and review. Feature requests, bug reports, and pull requests do not require sponsorship.

---

Thanks

• DeepSeek — thank you for the models and support that power every turn. 感谢 DeepSeek 提供模型与支持，让每一次交互成为可能。
• DataWhale 🐋 — thank you for your support and for welcoming us into the Whale Brother family. 感谢 DataWhale 的支持，并欢迎我们加入“鲸兄弟”大家庭。
• OpenWarp — thank you for prioritizing codewhale support and for collaborating on a better terminal-agent experience.
• Open Design — thank you for support and collaboration around design-forward agent workflows.

This project ships with help from a growing community of contributors:

• merchloubna70-dot — 28 PRs spanning features, fixes, and VS Code extension scaffolding (#645–#681)
• WyxBUPT-22 — Markdown rendering for tables, bold/italic, and horizontal rules (#579)
• loongmiaow-pixel — Windows + China install documentation (#578)
• 20bytes — User memory docs and help polish (#569)
• staryxchen — glibc compatibility preflight (#556)
• Vishnu1837 — glibc compatibility improvements and terminal restoration on SIGINT/SIGTERM (#565, #1586)
• shentoumengxin — Shell cwd boundary validation (#524)
• toi500 — Windows paste fix report
• xsstomy — Terminal startup repaint report
• melody0709 — Slash-prefix Enter activation report
• lloydzhou and jeoor — Compaction cost reports and npm installer stream-pause race fix (#1860); lloydzhou also contributed deterministic environment context (#813, #922) and KV prefix-cache stabilisation (#1080)
• Agent-Skill-007 — README clarity pass (#685)
• woyxiang — Windows install documentation (#696)
• wangfeng — Pricing/discount info update (#692)
• zichen0116 — CODE_OF_CONDUCT.md (#686)
• dfwqdyl-ui — model ID case-sensitivity compatibility report (#729)
• Oliver-ZPLiu — stale working... state bug report, Windows clipboard fallback, MCP Streamable HTTP session fixes, and Homebrew tap automation (#738, #850, #1643, #1631)
• reidliu41 — resume hint, workspace trust persistence, Ollama provider support, thinking-block stream finalization, CI cache hardening, streaming wrap, DeepSeek model completions, help picker selection polish, transcript user-message highlighting, approval one-step confirmation flow, and model-picker Esc-selection fix (#863, #870, #921, #1078, #1603, #1628, #1601, #1964, #1995, #2143, #2056)
• cyq1017 — Unicode git_status paths, local/configured skill discovery, mode-switch toast dedupe, sub-agent completion handoff compatibility, and goal-prompt actionability (#1953, #1956, #1957, #2057, #2120, #2097)
• xieshutao — plain Markdown skill fallback (#869)
• GK012 — npm wrapper --version fallback (#885)
• y0sif — parent turn-loop wakeup after direct child sub-agent completion (#901)
• mac119 and leo119 — codewhale update command documentation (#838, #917)
• dumbjack / 浩淼的mac — command-safety null-byte hardening (#706, #918)
• macworkers — fork confirmation with the new session id (#600, #919)
• zero and zerx-lab — notification condition config and richer OSC 9 notification body (#820, #920)
• chnjames — cached @mention completions, config recovery polish, and Windows UTF-8 shell output (#849, #927, #982, #1018)
• angziii — config safety, async cleanup, Docker hardening, and command-safety fixes (#822, #824, #827, #831, #833, #835, #837)
• elowen53 — UTF-8 decoding and deterministic test coverage (#825, #840)
• wdw8276 — /rename command for custom session titles and composer session-title display fix (#836, #2108)
• banqii — .cursor/skills discovery path support (#817)
• junskyeed — dynamic max_tokens calculation for API requests (#826)
• Hafeez Pizofreude — SSRF protection in fetch_url and Star History chart
• Unic (YuniqueUnic) — Schema-driven config UI (TUI + web)
• Jason — SSRF security hardening
• axobase001 — snapshot orphan cleanup, npm install guards, session telemetry fixes, model-scope cache clear, symlinked skill support, npm mirror-escape-hatch guidance, and proxy preservation for child tasks (#975, #1032, #1047, #1049, #1052, #1019, #1051, #1056, #1608)
• MengZ-super — /theme command foundation and SSE gzip/brotli decompression (#1057, #1061)
• DI-HUO-MING-YI — Plan-mode read-only sandbox safety fix (#1077)
• bevis-wong — precise paste-Enter auto-submit reproducer (#1073)
• Duducoco and AlphaGogoo — skills slash-menu and /skills coverage fix (#1068, #1083)
• ArronAI007 — window-resize artifact fix for macOS Terminal.app and ConHost (#993)
• THINKER-ONLY — OpenRouter and custom-endpoint model-ID preservation (#1066)
• Jefsky — DeepSeek endpoint correction report (#1079, #1084)
• wlon — NVIDIA NIM provider API-key preference diagnosis (#1081)
• Horace Liu — Nix package support and install documentation (#1173)
• jieshu666 — terminal repaint flicker reduction (#1563)
• gordonlu — Windows Enter / CSI-u input fix (#1612)
• mdrkrg — first-run onboarding crash fix when the API key is missing (#1598)
• Aitensa — CJK wrapping propagation for diff and pager output (#1622)
• qiyan233 — legacy DeepSeek CN provider alias compatibility (#1645)
• zlh124 — WSL2/headless startup report, clipboard-init fix, and YAML block-scalar frontmatter parsing (#1772, #1773, #1908, #1907)
• aboimpinto — Windows alt-screen logging, Home/End composer, and runtime log follow-ups (#1774, #1776, #1748, #1749, #1782, #1783)
• LeoLin990405 — provider model passthrough, reasoning replay, thinking-only turn, and Windows quoting fixes (#1740, #1743, #1742, #1744)
• nightt5879 — Ctrl+C prompt restore fix (#1764)
• h3c-hexin — streaming batch tool-call preservation, CLI reasoning-effort passthrough, sub-agent completion handoff compatibility, and self-hosted context budgeting (#1686, #1511, #2057, #2120, #2060)
• hxy91819 — prefix-cache preservation during tool-result pruning (#1514)
• JiarenWang — Plan-mode read-only enforcement, approval-takeover clamping, Ctrl+H delete fix, and undo context sync (#1123, #962, #958, #1150)
• Liu-Vince — MCP pagination, markdown indentation preservation, zh-Hans i18n polish, and env-var documentation (#1256, #1179, #1274, #1178)
• linzhiqin2003 — --model auto cost-saving bias, execution-discipline prompts, and declarative-fact memory hygiene (#1385, #1384, #1381)
• lbcheng888 — cost persistence across save/restore and transcript scroll fix (#1192, #1211)
• pengyou200902 — UTF-8-safe memory truncation, truncation-marker precision, and keybinding docs (#968, #1122, #1095)
• ChaceLyee2101 — reasoning-token cost tracking with auto-CNY on zh-Hans and zh-CN README sync (#1505, #1504)
• CrepuscularIRIS — low-motion mode for Termius/SSH and npx MCP server sandbox fix (#1479, #1346)
• laoye2020 — Catppuccin, Tokyo Night, Dracula, and Gruvbox themes with /theme picker (#1534)
• punkcanyang — Kitty (OSC 99) and Ghostty (OSC 777) desktop notification support (#1426)
• Rene-Kuhm — Spanish (es-419) Latin American localization (#1452)
• sternelee — DeepSeek prefix-cache stability tracking (#1517)
• ComeFromTheMars — Shift+Up/Down transcript scroll shortcuts (#1432)
• sockerch — pinyin aliases for all slash commands (#1306)
• Apeiron0w0 — FocusGained debounce for Tabby terminal flicker loop (#1560)
• greyfreedom — jump-to-latest-transcript button (#969)
• SamhandsomeLee — explicit hidden-file mention completion (#1270)
• dst1213 — quota-error HTTP 400 retry (#1203)
• fuleinist — --yolo flag forwarding from CLI to TUI (#1233)
• heloanc — Home/End key composer support (#1246)
• jinpengxuan — active provider credential preservation during onboarding (#1265)
• lixiasky-back — verified npm binary adoption (#1339)
• J3y0r — workspace-switch command (#1065)
• KhalidAlnujaidi — delegate skill bundling (#1144)
• Wenjunyun123 — docs anchor-offset preservation (#1282)
• whtis — zh-CN README dispatcher-path sync (#1235)
• aqilaziz — memory skill-link fix (#1095)
• wuwuzhijing — rsproxy rustup workaround install docs (#1011)
• eltociear — Japanese README translation (#746)
• Ling — grep_files cancellation-token support and Ctrl+Z composer-draft recovery (#1839, #1911)
• Ben Younes — Linux Wayland (non-wlroots) clipboard support (#1938)
• Matt Van Horn — Docker first-run permission fix and runtime system-prompt regression tests (#1699, #1702)
• Kristopher Clark — compaction user-query preservation fix (#1704)
• tdccccc — composer scroll fix and pager mouse-wheel support (#1715, #1716)
• LittleBlacky — provider-gated reasoning_content stream fix (#1680)
• Anaheim — rlm_open blank-source schema validation report (#1712)
• THatch26 — terminal resize paging fix (#1724)
• Alvin — Zed ACP id compatibility report (#1696)
• knqiufan — sub-agent file-write delegation work (#1833)
• IIzzaya — exact-alias-first slash-completion ordering idea (#1811)
• DC — terminal cleanup-guard idea (#1630)
• imkingjh999 — provider/model switching fixes (#1642)
• Photo — provider-aware /model picker catalog work (#1201)
• chennest — diagnostics schema report (#1685)
• kunpeng-ai-lab — Windows composer scroll fix (#1578)
• WuMing — Windows PowerShell flicker fix (#1591)
• maker316 — LoopGuard/checklist loop report (#1574)
• lalala — approval denial regression report (#1617)
• muyuliyan — pandoc_convert validation fix (#1523)
• czf0718 — resize and turn-completion flicker fix (#1537)
• MeAiRobot — toast overlay composer-input fix (#1485)
• tiger-dog — approval modal collapse and markdown identifier fixes (#1455)
• MMMarcinho — opt-in image_analyze vision tool (#1467)
• lucaszhu-hue — AtlasCloud provider integration (#1436)
• sandofree — Tavily and Bocha web_search backends (#1294)
• zhuangbiaowei — /change release-notes command (#1416)
• NorethSea — updater companion-binary refresh fix (#1492)
• Jianfengwu2024 — Windows MSVC toolchain environment preservation (#1487)
• Fire-dtx — npm postinstall recoverability work (#1059)
• oooyuy92 — long-session palette readability report (#1070, #936)
• qinxianyuzou — zh-Hans destructive approval wording (#1087, #1091)
• tyouter — session title/history preview cleanup (#1510)
• xulongzhe — issue-template and vision-boundary follow-ups (#1530, #1544)
• YaYII — trusted media path work (#1462)
• 47Cid and Jafar Akhondali — responsible security disclosures and hardening reports
• gaord — approval-remember live-turn sync fix and user-message transcript highlighting (#2041, #2047)
• idling11 — readable /restore snapshot labels and sidebar hover tooltips (#2111, #2110)

---

Contributing

See CONTRIBUTING.md. Pull requests welcome — check the open issues for good first contributions.

If you want CodeWhale to help improve CodeWhale, start with the recursive self-improvement prompt. It is designed to turn one DeepSeek V4 Pro session, or another capable open-weight path, into one small, reviewable patch.

Note

Not affiliated with DeepSeek Inc.

License

MIT

Star History