DeepSeek Harness Plugin Hub

Publish and manage complete Harness Profiles. Discover Plugins for your next setup.

Explore

PluginsPresetsDocsNews

Community

Publish a pluginContactReport an issue

Resources

Plugin Hub on GitHubDeepSeek HarnessSystem statusPrivacy notice
© 2026 DeepSeek Harness Plugin HubPowered byPaxTech

Independent and unofficial. Not affiliated with, authorized by, or endorsed by DeepSeek.

DeepSeek Harness Plugin Hub
ProfilesPluginsCategoriesNewsDocsSign inManage Profiles
ProfilesPluginsCategoriesNewsDocsSign in
← Plugins
M

@deepseek-ai/dsh-model-audit

Model Audit

Independent local DSH plugin for request and response model metadata; not an official DeepSeek plugin

The plugin will be installed here. Keep web if you are unsure.

npx -y @deepseek-ai/dsh plugin --profile web add github:jghwwnq/dsh-model-audit#982d77402df30e10be4d2010250cfecf9a558a88
READMECompatibilityVersions

Compatibility and provenance

Model Audit is published as @deepseek-ai/dsh-model-audit and currently resolves to version 0.1.7-rc.2. The Hub verifies its manifest and preserves the exact installation source for reproducible installs.

DSH compatibility
*
Runtime surfaces
web
Release source
github
Registry updated
9/25/2026

Versions

0.1.7-rc.2prerelease
9/25/2026

Related plugins

Loading related plugins…

Latest
0.1.7-rc.2
DSH
*
HMR
Process restart
Tree shaking
Safe tree shaking not declared
Unpacked size
Unavailable
Files
Unavailable
Surface
web
License
MIT
Source
github
GitHub
★ 1
Weekly downloads
0
Last push
9/25/2026
View source ↗
README badge

Click the badge to copy Markdown for your README.

Do you maintain this Plugin?Claim benefit · Priority security scan

Verify the GitHub repository declared in package.json to manage this listing. After you claim it, Hub will prioritize a security scan of the current version and publish the result when it passes.

Claim this Plugin →
Report an issue
Model Audit — DSH Plugin for DeepSeek Harness

Related plugins

More verified plugins in models-usage.

Usage@linxin666/dsh-usageUsage statistics plugin for the dsh web GUI: per-provider balance and coding-plan quota detection plus a live token usage ledger, with a compact price glance card above the sidebar's Settings rowWhale Widgetdsh-whale-widgetDeepSeek balance whale widget in the bottom-right corner of the DSH Web interface: balance/today’s usage/peak-off-peak pricing, customizable bubble click sequence (text/balance/today/peak-off-peak/image/random phrases and parallel weighted selection), per-line styles and fonts, floating quick editinCommandcode Provider@mars-sea/dsh-commandcode-providerUnofficial DeepSeek Harness LLM provider plugin for Command Code, ported from pi-commandcode-provider (MIT). Registers the 'commandcode' provider route with a Models-page card and live model catalog.Plugin Subscriptionsdsh-plugin-subscriptionsUse ChatGPT (Codex), Claude, Grok (X Premium), GitHub Copilot, and Google Antigravity subscriptions as DeepSeek Harness LLM providers, with OAuth login from the web Settings page

README

DSH Model Audit — independent local plugin

中文安装与维护说明

An unofficial, independently built and installed plugin for DeepSeek Harness. It compares the actual outgoing request model with the upstream response model for DeepSeek Messages, GPT Responses and Chat Completions, in both the Web version and the desktop application.

Not affiliated with or endorsed by DeepSeek. The @deepseek-ai/dsh-model-audit package ID is kept from the original local install for configuration migration; this package is not published to npm, and the name is local-only — do not publish it under DeepSeek's npm scope. Licensed under MIT.

Install the directory or packed tarball with DSH's plugin manager, or:

dsh plugin --profile web add /absolute/path/to/dsh-model-audit --offline --ignore-scripts

The desktop application owns its profile exclusively, so the CLI refuses --profile desktop; install there from the application's own plugin page (Installed plugins → add the absolute directory path) instead.

The package appears under Installed plugins. After the first external-directory installation, restart DSH once so the host registers that linked package's peer resolution. A path change from an already loaded version also requires a restart. Then enable Model monitor, open its detail panel and select Open monitor; subsequent on/off switches work live. Records are held only in memory and are cleared on disable/restart.

Open monitor behaves per surface: the Web version opens the dashboard in a new browser tab, while the desktop application renders it inside the application window, because that shell denies every popup on its own dsh-app: scheme. The desktop view fetches the same authenticated document and renders it in a same-origin frame; close it with the header button or Escape. Host-side plugin changes still need a desktop restart, because an already imported module stays cached for the life of that process.

The package contains built host/browser artifacts. It does not edit the official repository, workspace, lockfile, built-in roster, or user credentials. All runtime framework imports use DSH's host-supplied peer dependencies. Plugin version 0.1.7-rc.2 targets and is tested against DSH 0.1.7-rc.2; future host API changes may require a plugin update.

To upgrade from 0.1.7-rc.1, install the updated directory or deepseek-ai-dsh-model-audit-0.1.7-rc.2.tgz, then reload DSH so it reads the new package version and peers. This release passes the native compatibility check without a version exemption; 0.1.7-rc.1 stopped passing it because DSH 0.1.7-rc.2 split the Messages adapter config, and the plugin's bundled test fixture now loads @deepseek-ai/dsh-llm-deepseek-api-key for the API-key provider. Keep the older plugin package when using the older DSH release.

Only calls through llm/stream and global fetch are observed. WebSocket, private transports and replay are reported as unobserved. The dashboard retains model/response identifiers, usage, timing and call-purpose metadata, not prompts, completions or credentials. An upstream model-name declaration cannot attest model weights or compute allocation.

For development, link to a built, unmodified DSH checkout without modifying it:

node scripts/link-dev.mjs /absolute/path/to/deepseek-harness
npm run build
npm test
DSH_SOURCE=/absolute/path/to/deepseek-harness npm run test:web
npm pack --ignore-scripts

The web suite uses the stock host's test scaffold via the explicit DSH_SOURCE development setting. Runtime use does not need that setting. Bundled parser licensing is in THIRD_PARTY_NOTICES.txt.

Repository layout

  • src/ — host half (index.ts, monitor.ts, parser.ts, dashboard.ts, locales.ts) and browser half (src/client/).
  • lib/ — committed build output (lib/index.js, lib/client.js and their types) that DSH loads; npm run build regenerates it.
  • tests/ — monitor unit tests, a real Loader composition over the stock DeepSeek/GPT adapters, and a browser integration suite; tests/expected/ holds their expected output.
  • scripts/link-dev.mjs — links this plugin's development dependencies to an already-built DSH checkout without modifying it.
  • dist/ — untracked local npm pack artifacts.

Supported host versions are exact: this release targets DSH 0.1.7-rc.2 and DSH refuses an incompatible plugin at load rather than risking a crash, so a newer DSH prerelease may need a matching plugin update. MIT license; the only third-party code inside the artifacts is eventsource-parser, and every framework import resolves to the host's own DSH packages at runtime.