CtrlK
BlogDocsLog inGet started
Tessl Logo

desktop-brand-builder

Generate a branded Qwen Code desktop package from the Tauri desktop shell using a minimal brandId and logo. Use when the user wants a custom, white-label, or rebranded desktop client, installer, DMG/EXE/AppImage/deb, or one-click brand build on top of packages/desktop-shell.

72

Quality

88%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

SecuritybySnyk

Passed

No findings from the security scan

SKILL.md
Quality
Evals
Security

Desktop Brand Builder (Tauri shell)

Goal

Create a branded desktop package from packages/desktop-shell with the least user input possible. The user should usually provide only:

brandId: acme-ai
logo: /absolute/path/to/logo.png
website: https://acme.ai

website is optional. Do not ask for app name, app id, artifact name, copyright, or updater endpoints unless the user explicitly asks to override them.

This skill replaces the Electron-era brand builder that lived in the removed packages/desktop. The Tauri shell is the only desktop implementation now; branding hooks are src-tauri/tauri.conf.json, src-tauri/icons/, and the bootstrap/ startup UI.

Input Rules

Required fields:

  • brandId: must match ^[a-z][a-z0-9-]*$
  • logo: local file path; must exist; .png recommended (square, >= 1024px)

Optional overrides:

  • website
  • appName
  • appId (Tauri bundle identifier)
  • artifactPrefix
  • updaterEndpoints (JSON array; empty array disables in-app updates)
  • updaterPubkey (base64 public key; required when updaterEndpoints is non-empty — must match the TAURI_SIGNING_PRIVATE_KEY used to sign your updater artifacts)
  • target: mac, win, linux, or all

If required input is missing, ask once:

请提供:
brandId: 例如 acme-ai,只能小写字母、数字、短横线
logo: 本地 logo 文件路径(建议 1024x1024 PNG)
website: 可选

Once the required fields are present, proceed without a confirmation step.

Derived Defaults

Infer missing values deterministically:

  • appName: title-case the hyphen-separated brandId; acme-ai becomes Acme AI
  • artifactPrefix: title-case the hyphen-separated brandId and join with hyphens; acme-ai becomes Acme-AI
  • appId: if website has a valid host, reverse the host labels and append .desktop; https://acme.ai becomes ai.acme.desktop
  • fallback appId: app.<brandId>.desktop
  • updaterEndpoints: empty by default. A branded build must never poll the official Qwen Code updater feed, and the official feed must never update a branded build. Only set endpoints when the user supplies their own feed.

Workflow

Work in an isolated build clone so the working repository stays clean:

BUILD_ROOT="$PWD/brand-builds/<brandId>-<timestamp>"
mkdir -p "$BUILD_ROOT"
git clone --branch main --single-branch \
  https://github.com/QwenLM/qwen-code.git \
  "$BUILD_ROOT/qwen-code"
cd "$BUILD_ROOT/qwen-code"
git checkout -B brand-<brandId> origin/main

If the clone or checkout fails, stop and report the failure. Do not continue as if brand-<brandId> was created.

Create a temporary brand.json in the build directory:

{
  "brandId": "acme-ai",
  "logo": "/absolute/path/to/logo.png",
  "website": "https://acme.ai",
  "appName": "Acme AI",
  "appId": "ai.acme.desktop",
  "artifactPrefix": "Acme-AI",
  "updaterEndpoints": [],
  "updaterPubkey": ""
}

Install dependencies. The brand script itself only needs desktop-shell's own node_modules, but npm run build:runtime shells out to the repo root (which uses cross-env and other root devDependencies), so the root install is also required before packaging:

# Root dependencies (needed by build:runtime → cross-env, esbuild, etc.)
npm install

# Desktop-shell dependencies
cd packages/desktop-shell
npm install --workspaces=false
cd ../..

Then run this skill's bundled brand creation script with plain Node (the script has no dependencies beyond Node >= 18):

node packages/desktop-shell/.agents/skills/desktop-brand-builder/scripts/brand-create.mjs \
  --shell-root /absolute/path/to/qwen-code/packages/desktop-shell \
  --config /absolute/path/to/brand.json

The agent should not hand-edit tauri.conf.json, icon files, or bootstrap brand strings when this bundled script is available. The bundled script is the source of truth for patching config and generating resources.

What the script does:

  1. Patches src-tauri/tauri.conf.json: productName, identifier, bundle.shortDescription, and plugins.updater.endpoints. When updaterEndpoints is empty it also clears bundle.createUpdaterArtifacts and blanks the official plugins.updater.pubkey (set to empty string rather than deleted, because the updater plugin requires the field); a brand supplying its own feed must supply its own pubkey.
  2. Regenerates the full icon set from the logo via npx --yes @tauri-apps/cli icon <logo> (falls back to a warning if the CLI cannot run; in that case copy the logo over src-tauri/icons/icon.png manually and tell the user the remaining sizes are stale).
  3. Patches the bootstrap UI: page title, brand heading, startup strings in bootstrap/index.html and bootstrap/bootstrap.js, and replaces bootstrap/qwen-code-logo.svg usage with the brand logo.

Package with the current host target unless the user requested a target:

cd packages/desktop-shell
npm run build:runtime --workspaces=false
npx tauri build            # current platform

Cross-compile: build:runtime bundles the Node runtime for the platform indicated by QWEN_DESKTOP_TARGET (defaults to the host). When targeting a different platform you must re-run build:runtime with the env var set before each tauri build --target, otherwise the packaged artifact contains a wrong-arch Node binary and fails at launch with an exec format error:

# Cross-compile: set QWEN_DESKTOP_TARGET and re-run build:runtime per target
QWEN_DESKTOP_TARGET=aarch64-apple-darwin npm run build:runtime --workspaces=false
npx tauri build --target aarch64-apple-darwin   # explicit macOS arm64

For target: all, iterate build:runtimetauri build per target; run only targets supported by the current machine or CI environment. Do not claim cross-platform artifacts were produced unless the files exist. Artifacts land under packages/desktop-shell/src-tauri/target/release/bundle/ for the host target, or src-tauri/target/<triple>/release/bundle/ when --target <triple> is used.

Signing and Updates

Branded builds are unsigned by default. The upstream release pipeline's signing secrets (Apple, Windows) and updater private key belong to the official Qwen Code releases only. For a brand that needs signed releases or in-app updates, set up separate credentials and a separate updater feed; do not reuse the upstream ones.

To generate a signing key pair for your updater feed:

npx @tauri-apps/cli signer generate -w ~/.tauri/my-brand.key
# The .key file is the private key (set as TAURI_SIGNING_PRIVATE_KEY in
# your build CI). The corresponding .pub file contains the base64 public
# key — paste it into brand.json as updaterPubkey.

Validation

After packaging:

  1. Confirm the expected artifact exists under packages/desktop-shell/src-tauri/target/release/bundle/ (or src-tauri/target/<triple>/release/bundle/ for cross-compile targets) (dmg/, nsis/, appimage/, or deb/).
  2. Compute sha256sum or shasum -a 256 for each artifact.
  3. On macOS, run hdiutil verify for generated DMG files.
  4. Report the artifact path, SHA-256, app name, app id, and build directory.

Failure Handling

  • Invalid brandId: show the regex and ask for a corrected value.
  • Missing logo: ask for a valid local path.
  • Missing bundled script: report that packages/desktop-shell/.agents/skills/desktop-brand-builder/scripts/brand-create.mjs is missing, and include the expected command.
  • Already-branded shell-root: the script refuses to run when productName is no longer the default (Qwen Code Desktop). Start from a fresh clone — do not re-run the script in an already-patched tree.
  • Build failure: preserve the build directory, return the last useful error lines, and include the full log path or command that produced the failure.

Do not delete the build directory on failure. Never re-run brand-create in the same clone — the script is single-use. The "preserve the build directory" guidance is for post-mortem debugging, not for retrying the brand step. If the brand config was wrong, discard the clone and start fresh.

Repository
QwenLM/qwen-code
Last updated
First committed

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.