Back to directory

dsh-composer-assist

Curated pickMaintenance: Active

lphehe/dsh-composer-assist

Suggest a next message, complete a draft or polish a prompt from the current session context.

View on GitHubHomepage
$ dsh plugin --profile web add https://github.com/Lphehe/dsh-composer-assist/releases/download/v0.1.1/dsh-composer-assist.tgz

17

stars

2

forks

JavaScript

Language

NOASSERTION

License

2026-10-22

Created

2026-10-24

Last push

Predict a follow-up, complete a draft or polish a prompt in DSH. Tab only inserts the draft. Real model calls may incur costs; source modification requires prior written permission.

DSH integration

Native runtime

Author-claimed

Safety audit

Unaudited

Last verified

2026-10-24

License

Not specified by the author

01What can it help you accomplish?

02How to install into DeepSeek Harness

Prerequisites

  • Configure an available model in DSH; when no prediction model is set, the plugin follows the session model last used.

Installation steps

  1. 01

    $ dsh plugin --profile web add https://github.com/Lphehe/dsh-composer-assist/releases/download/v0.1.1/dsh-composer-assist.tgz

Verify the integration

Not specified by the author

03DSH integration and capability boundaries

DSH integrationNative runtime

DSH composer prediction, completion and polishing plugin

04Who is it for? When not to use it?

Good for

  • DSH users composing a follow-up prompt

Not for

  • Automatic mode predicts once after each answer using a real billed model call. Select manual-only or disable it to avoid automatic calls.
  • Source is available under restrictive terms, not an open-source license. Modification, derivative versions and distribution of modified versions require prior written permission.
  • Model requests include recent dialogue, user constraints, bounded tool text, filtered top-level workspace filenames and a Git status summary.
  • Upstream cancellation is best effort and may not immediately stop generation or billing.

05Compatibility, maintenance and safety notes

  • Automatic mode predicts once after each answer using a real billed model call. Select manual-only or disable it to avoid automatic calls.
  • Source is available under restrictive terms, not an open-source license. Modification, derivative versions and distribution of modified versions require prior written permission.
  • Model requests include recent dialogue, user constraints, bounded tool text, filtered top-level workspace filenames and a Git status summary.
  • Upstream cancellation is best effort and may not immediately stop generation or billing.
2026-10-222026-10-24v0.1.1

README author claims; not independently audited.

06Frequently asked questions

Will accepting a suggestion send a DSH message?

No. Tab or Accept only inserts it into the draft; you decide whether to send.

Will this make additional DSH model calls?

Yes. Automatic mode makes a real billed prediction after each answer; manual-only mode is available.

Which DSH model is used?

If no prediction model is selected, it follows the session model last used. Configure a model in DSH if none is available.

Can I modify this DSH plugin?

The README requires prior written permission for modifications or derivative versions.

08Data and sources

  • Author-claimedgithub.comce104ff92678…

    在 DSH 中打开 **插件 → 添加插件**

  • Author-claimedgithub.comce104ff92678…

    结合当前对话预测后续输入、补全草稿,或按需润色提示词。建议按会话保留;按 **Tab** 只填入草稿,**不会自动发送**。

  • Author-claimedgithub.comce104ff92678…

    DSH **0.2.1-rc.2**

This page is generated from the project’s public documentation, repository metadata and a structured parse of DSH Plugins; last verified on 2026-10-24. Found an error? Submit a correction.

🏆

Best DeepSeek Harness Plugins

Twelve plugins worth installing first — picked from the whole catalog, across every category.

AD

DSH Plugins is an independent community directory of DeepSeek Harness plugins. Not affiliated with or endorsed by DeepSeek. Third-party plugins are not security-audited — review the source before installing.

New DeepSeek Harness plugins, weekly. No spam.