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

dsh-model-key-sync

Model Key Sync

Sync your DeepSeek Harness model configuration and third-party API keys across machines, over S3-compatible storage or WebDAV.

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

npx -y @deepseek-ai/dsh plugin --profile web add github:tesths/dsh-model-key-sync#008bc076772a211b6788187f913099b79665f87a
READMECompatibilityVersions

Compatibility and provenance

Model Key Sync is published as dsh-model-key-sync and currently resolves to version 1.2.0. The Hub verifies its manifest and preserves the exact installation source for reproducible installs.

DSH compatibility
*
Runtime surfaces
web
Release source
github
Registry updated
10/18/2026

Versions

1.2.0stable
10/18/2026
1.1.0stable
10/18/2026

Related plugins

Loading related plugins…

Latest
1.2.0
DSH
*
HMR
Process restart
Tree shaking
Safe tree shaking not declared
Unpacked size
Unavailable
Files
Unavailable
Surface
web
License
MIT
Source
github
GitHub
★ 0
Weekly downloads
0
Last push
10/18/2026
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

Related plugins

More verified plugins in security-access.

Experimental Auto Review@deepseek-ai/dsh-experimental-auto-reviewPer-tool LLM authorization review for the DeepSeek Harness Auto permission presetCc Safety Netcc-safety-netA coding agent CLI hook - block destructive commands and secret file accessDoctor@linxin666/dsh-doctorTransactional rescue mode for DSH profiles with a supervised launcher, isolated recovery capsule, deterministic repairs, health monitoring, and a local Web recovery consolePocketdsh-pocketPut DeepSeek Harness in your pocket: one package, one settings page, and scan a QR code on your phone to access DSH on your computer in sync (LAN + public network, real-time screen mirroring).

README

dsh-model-key-sync

Sync your DeepSeek Harness model configuration and third-party API keys across machines, over S3-compatible storage or WebDAV.

Sign in to Harness on a new computer, pull once, and your custom providers and their keys are there — no retyping one key per provider.

Keys are stored in the remote as plaintext

This is deliberate — it matches CC Switch, and it means a new machine needs no passphrase. The consequence is blunt: anyone who can read that bucket or collection can read every API key in it. Use a dedicated application password, keep the cloud account itself safe, and turn the key toggle off if you only want model declarations to travel. See Security.

Why this exists

Harness keeps each machine's profile, credentials and model declarations locally. Moving to a second machine normally means re-entering every provider: display name, endpoint, protocol, model catalog, and the key.

Other plugins in the ecosystem sync the profile directory and local plugin sources — dsh-cloud-sync documents that it deliberately does not copy plaintext credentials. That leaves exactly this gap: the keys. This plugin fills it, and carries the model declarations that reference them.

Install

Ask the agent in Harness:

Install the plugin from github:tesths/dsh-model-key-sync

It runs the bundle installer for you. Or point it at a local checkout:

Install the plugin at /path/to/dsh-model-key-sync

Then reload the page. The section appears under Settings → Model Key Sync.

Requires a Harness build that provides the settings, credentials and tools services, and an llm-pi-ai (or compatible) model adapter row.

Configure

Settings → Model Key Sync. Cloud sync is the page; a read-only summary of this machine — display name, API endpoint and model count — sits below it.

WebDAV

FieldValue
BackendWebDAV
Collection URLhttps://dav.jianguoyun.com/dav/ (nutstore), or your Nextcloud path
User namethe account's login name (for nutstore: your email)
Passwordan application password, not the login password
Remote directorydsh-model-key-sync
Sync groupany name; identical on every machine

S3-compatible

R2, OSS, B2, MinIO and AWS all work. Fill bucket, region (auto for R2), endpoint (blank for AWS), access key id and secret.

Fill in the fields and press Save (configuration and secret save together), then Test → Upload.

On the second machine

Install the plugin, open the same page, fill the same values including the sync group, and press Pull. The status card lists the credentials it stored, for example:

pulled
credentials: ACME_API_KEY, OTHER_API_KEY

Daily use

  • Manual: upload after editing models under Settings → Models; pull before using another machine.
  • Auto: tick Auto upload. A pull never triggers an upload.
  • Conflict: nothing is overwritten silently. When both sides changed, a pull is refused and the message names the axis — model routes, API keys, or both. Tick Force to take the remote, or upload to keep this device.

What travels

IncludedNot included
Provider routes: display name, protocol, endpoint, model catalog, key reference, provider defaults (reasoning, input, cache, thinking budgets)Sessions, attachments, plugin sources
The API key each route's apiKeyEnv resolves toCredits/quota, usage history
Your choice of cloud backend and sync groupOther settings (theme, shortcuts, …)

Restoring merges by route name and preserves fields it did not export — retry policy, timeouts, transport or compatibility switches stay as they are on this machine.

Remote layout

<remoteRoot>/v1/<syncGroup>/
├── manifest.json        format, version, device, per-object sha256 and size, snapshotId
├── <provider>.json      one file per route
└── keys.json            { "ENV_NAME": "value" }

Payload objects are written first and the manifest last, so a reader never sees a manifest describing objects the writer has not finished uploading. Every object is verified by size and SHA-256 on download. Transfers are serialized with a total timeout, an idle timeout and a response-size cap.

Security

  • Keys leave the machine only through ctx.credentials.resolve and are stored through ctx.credentials.set — the same seam the model adapter reads.
  • They are uploaded unencrypted. Treat the remote as a secret store: a private bucket, an application password you can revoke, no shared folders.
  • Turn off Sync keys to sync model declarations only.
  • Anyone who obtains the remote can use your keys. Revoke them at the provider if that happens.
  • The local sync state (~/.dsh/model-key-sync/cloud-state.json) stores hashes, not keys.

Development

npm install
npm test                 # node test/host.mjs && node test/client.mjs
node test/host.mjs       # 116 checks: settings, files, command channel, keys, S3 + WebDAV
node test/client.mjs     # 91 checks: one-page cloud sync, read-only local summary, command channel

test/host.mjs starts a real S3-compatible server and a real WebDAV server on loopback and drives two simulated devices through upload, pull, idempotency, conflict refusal and forced overwrite. test/client.mjs loads the real client bundle into jsdom with a faithful configForms/locale/slots contract.

See docs/ARCHITECTURE.md for why the page talks to the Host the way it does, and CONTRIBUTING.md before changing that channel.

License

MIT