deepseekplugins

DeepSeek Harness plugin

dsh-reasoning-cn

DeepSeek Harness plugin that steers reasoning toward Simplified Chinese and translates eligible reasoning blocks.

Tools & Capabilities
Jump to install

Source facts

Repository
William123666/dsh-reasoning-cn
Latest update
Aug 19, 2026
Category
Tools & Capabilities
GitHub stars
1
Format
plugin
Catalog evidence
Upstream dsh.bundle evidence
Evidence path
package.json#dsh.bundle
Checked against
0.1.0-rc.8
Upstream check date
2026-08-20

This evidence comes from the upstream catalog. This site has not installed, run, or security-reviewed the plugin.

Install

Start with a prompt that asks an agent to review the GitHub repository and source. Switch to the command if you want to install it yourself.

Copy this prompt into DSH, Codex, or another agent and ask it to review the GitHub repository and source first.

Do not install or run any commands yet. Read this plugin's GitHub repository, README, and relevant source code. Then answer the questions below clearly and directly so I can decide whether it fits my needs:

1. What is this plugin, and what problem does it solve?
2. Who is it for, and what are its typical use cases?
3. How is it used after installation? Include one minimal example.
4. What known limitations or privacy, security, compatibility, or maintenance risks does it have?
5. Give a clear recommendation: recommend, conditionally recommend, or do not recommend, with reasons.

Distinguish statements documented by the repository, inferences from source code, and unknowns. If evidence is insufficient, say so explicitly. Do not guess or simply repeat the README.

GitHub: https://github.com/William123666/dsh-reasoning-cn
Plugin: dsh-reasoning-cn
Author: William123666

Check the source files

Read the README and other files from this plugin directory before installing.

File explorer4 files
README.en.mdSource · read only
README language

dsh-reasoning-cn

中文

DeepSeek Harness plugin that steers reasoning toward Simplified Chinese and can translate eligible reasoning blocks at the llm/stream seam.

> [!NOTE] > This is an unofficial community plugin. It is not affiliated with, maintained by, or endorsed by DeepSeek or the DeepSeek Harness maintainers.

> [!WARNING] > When enabled, the translation layer sends the complete reasoning block to the configured provider/model and replaces the reasoning stored in session logs, Web replay, and subsequent model context. Reasoning may contain user data, tool output, source code, paths, or accidental credentials. Translation is disabled by default; enable it only after checking the provider's retention policy, access scope, and cost.

Install

dsh plugin --profile web add dsh-reasoning-cn

The bundled Cordis patch mounts dsh-reasoning-cn. Set translate: true explicitly to enable translation. The steering layer remains available with translate: false and makes no auxiliary LLM calls.

Compatibility

  • Verified with @deepseek-ai/cordis 4.0.1, the DSH 0.1.0-rc.7 package set, Node.js 22.19.0, and current 24.x.
  • DeepSeek Harness is in Developer Preview and may make breaking API changes. Run npm run verify in a test environment before upgrading DSH, and include all versions in a compatibility report.
  • Peer dependencies intentionally remain pinned to the verified rc.7 release until a newer version has passed the suite.

Development

npm ci
npm run typecheck
npm test
npm run build
npm run smoke
npm run package:check
npm run verify

Requires Node.js 22.19 or newer. The development overlay in mount/dev.cordis.yml uses a placeholder absolute path; replace it before mounting a local build.

Configuration

FieldDefaultMeaning
steeringbothoff, system-section, first-step-reminder, or both
translatefalseMaster switch for auxiliary reasoning translation
translateProviderdeepseek-officialProvider route for translation calls
translateModeldeepseek-v4-flashTranslation model route
translateThinkingOfftrueRequests reasoningEffort: 'off' for translation
cjkThreshold0.3CJK ratio at or above which text counts as Chinese
maxCharsPerBlock20000Maximum buffered reasoning characters before verbatim fallback
timeoutMs30000Per-translation timeout in milliseconds
maxOutputTokens8192Maximum translation output tokens
verbosefalseLogs translation size and duration

For a DSH profile patch, configure the mounted plugin with a block such as:

- insert:
    - id: reasoning-cn
      name: dsh-reasoning-cn
      config:
        steering: both
        translate: true
        translateProvider: your-provider
        translateModel: your-translation-model
        translateThinkingOff: true

The translation layer sends complete eligible reasoning to the configured provider. Successful output replaces reasoning in session logs, Web replay, and subsequent model context; failed validation, timeout, or provider errors keep the original text. It is not a redaction layer and cannot fully prevent prompt injection. See the Chinese README for the complete behavior and limitations.

Real-provider verification (maintainers only)

npm run verify:real sends the test prompt and model reasoning to a live provider and can incur cost. It never runs in CI. Build first, then provide all route and credential-environment details explicitly:

VERIFY_PROVIDER=your-provider \
VERIFY_API_KEY_ENV=YOUR_PROVIDER_API_KEY \
VERIFY_BASE_URL=https://your-provider.example/v1 \
VERIFY_API=openai-completions \
VERIFY_MODEL=your-model \
npm run verify:real

Keep the actual credential only in the shell environment. The check disables steering, requires an auxiliary translation request, and verifies that the resulting reasoning passes the Chinese-content gate.

Add the dsh-plugin topic to the GitHub repository for DeepSeek Harness community discovery.

Releases

See [RELEASING.md](./RELEASING.md) for maintainer release policy, versioning, and npm Trusted Publishing prerequisites.

License

MIT, see [LICENSE](./LICENSE).

Back to searchOpen GitHub repository

More related plugins

View category