Mars-Sea/dsh-commandcode-provider
Unofficial Command Code LLM provider: registers a `commandcode` route with a live model catalog and reasoning-effort support.
Install
via npm
Resolves a published tarball rather than cloning the repository, and installs without any extra setup. Swap `web` for your profile name if you run another one.
via GitHub · npm package
Installing from GitHub runs the project's build script, which pnpm blocks until you allowlist it — run the command once and pnpm prints the exact key to add under `allowBuilds` in ~/.dsh/profiles/web/pnpm-workspace.yaml.
What happened when we ran it
Installed cleanly when we ran it
Every command here is run in a throwaway container against a clean profile, and the result is whatever the harness recorded — not a guess from the source. Last run 27d ago.
Show it in your README
For maintainers: the badge serves this listing's latest sandbox verdict, so a re-run updates it on its own — and it links readers to the full result here.
Due diligence
What it is — Integrates Command Code as a provider for DeepSeek Harness, registering a commandcode route with live model catalog and reasoning-effort support.
Who it is for — Users using Command Code models for reasoning tasks benefit from this. If your workflow already includes reasoning-effort through official CLI, you can skip it.
Watch out — Sandbox test passed: installed in a new profile and registered into the profile. No obvious issues found. Configuration applies to the next request without restart.
The verdict — I would install it because it provides reasoning-effort support and Image markers, but only if I have a Command Code subscription.
Generated by grok-4.6, and a starting point rather than a verdict. Where it says a plugin installs or does not, that is from a real run in a clean profile — everything else is read off the repository. Trust the source over this.
dsh-commandcode-provider is a DeepSeek Harness plugin that adds Command Code as an unofficial LLM provider. It registers a commandcode route and translates harness requests into Command Code’s Provider API at POST /alpha/generate. Its model list comes from GET {apiBase}/provider/v1/models and is cached in ~/.commandcode/models-cache.json. The plugin also adds Settings → Command Code, where the API base, working directory, request and stream timeouts, and API key can be configured. Credentials are stored through the DSH credentials service; connection settings are placed in the llm-commandcode section and apply to the next request.
dsh-commandcode-provider resolves credentials from config.apiKey, the apiKeyEnv credential reference, the launch environment, or ~/.commandcode/auth.json. It annotates models with plan, deal, image and context information, supports catalogued reasoning-effort levels, and sends images through the DSH attachment service when the selected model supports Vision.
It is intended for developers who already have a Command Code account and API key or subscription and want those models in an agent. It is the wrong choice where an official integration is required, where Command Code account setup is unavailable, or where the unofficial reverse-engineered API dependency is unacceptable. The official CLI requires Node.js 22 or newer; Command Code’s terms apply, and the project is not affiliated with Command Code, Inc.
The plugin exposes a Command Code settings page. Its connection fields are written to the llm-commandcode configuration section and take effect on the next request without a restart.
| Setting | Purpose |
|---|---|
apiKey |
Direct API key configuration |
apiKeyEnv |
Credential reference; defaults to COMMANDCODE_API_KEY |
| API base | Command Code API endpoint |
| Working directory | Working directory used for requests |
| Request timeout | Request timeout |
| Stream timeout | Streaming timeout |
The model catalogue is fetched from GET {apiBase}/provider/v1/models and cached at ~/.commandcode/models-cache.json.
The API-key lookup order is config.apiKey, the credential reference named by apiKeyEnv, the launch environment, then ~/.commandcode/auth.json. The last file is created by the official CLI after authentication.
The documented authentication path requires Node.js 22 or newer:
npm i -g command-code@latest installs the CLI.cmd login is used on macOS and Linux.cmdc login is used on native Windows.The browser-based login stores the key in ~/.commandcode/auth.json. A key can instead be created in Command Code Studio and entered in Settings → Command Code, or supplied through COMMANDCODE_API_KEY.
The commandcode route uses POST /alpha/generate. Its live catalogue displays each model’s minimum plan, an active deal or FREE, an Image marker for Vision models, and a context size such as 1M, 256K or 262K. Models are ordered by plan tier: Go, GOAT, Pro, then Provider/Max.
Reasoning-effort controls are available for models that the official catalogue marks with supported levels. Other reasoning models think automatically. Vision-capable models receive image input through the DSH attachment service; text-only models reject image input with UNSUPPORTED_CONTENT.
This is an unofficial community integration. Use requires a Command Code account and API key or subscription, and Command Code’s terms apply. The plugin is not affiliated with Command Code, Inc. Its documented external files are ~/.commandcode/models-cache.json and, when CLI authentication is used, ~/.commandcode/auth.json.
Written from the project's own documentation and kept in sync with it. Where the two disagree, the source is authoritative — read the README on GitHub
Same category
Use ChatGPT (Codex), Claude, and Grok subscriptions as DeepSeek Harness LLM providers, with Settings login, model catalogs, usage, plus image_generate, video_generate, and x_search tools.
Installed cleanly when we ran it
YuJunZhiXue
DeepSeek Harness 破甲：让所有模型都能破甲，不同模型可换不同提示词；默认提示词面向国模「小码酱」。Jailbreak for every model — swap prompts per model. 求 Star 收藏 ⭐
Installed cleanly when we ran it
NanmiCoder
Safe automatic permissions for DeepSeek Harness.
Installed cleanly when we ran it
Akimiya-z
Quality gate for AI/Codex-generated pull requests: blocks TODO leftovers, leaked secrets, sloppy commits and red CI before they reach main.
Installs, but not as a profile layer
HanaAyane
DSH适用的Codex风格的思考强度滑块，以及大肥鱼跑步滑块。Codex-style model and reasoning-effort slider for DeepSeek Harness
Installed cleanly when we ran it
volcengine
Registers Ark Agent Plan, Coding Plan and postpaid model routes in the native DSH model picker.
No one-line install — this plugin lives inside a larger repository and publishes no npm package.