Parallel Search for DeepSeek Harness

This plugin makes DeepSeek Harness's built-in web_search tool use Parallel Search. It does not add a new tool or provide web_fetch.

Quick start

You will need:

• Node.js 22.19 or later in the 22.x series, or Node.js 24 or newer;
• pnpm 10 or newer;
• a DeepSeek API key; and
• a Parallel API key.

Check your installed versions:

node --version
pnpm --version

If pnpm is missing, install it:

npm install --global pnpm@10

1. Get DeepSeek Harness

Run the supported Harness version:

npx --yes @deepseek-ai/[email protected] --version

You should see 0.1.0-rc.6. You do not need to install dsh globally. Keep using the complete npx commands shown below.

DeepSeek Harness is still a developer preview, so later versions may need a plugin update.

2. Install the Parallel plugin

Install the plugin into the web profile:

npx --yes @deepseek-ai/[email protected] \
  plugin --profile web add \
  @parallel-web/dsh-web-search@rc

The @rc suffix installs the current release candidate. After version 0.1.0 is stable, you can leave the suffix off.

3. Start DeepSeek Harness

Make your Parallel API key available in the terminal where you will run Harness:

export PARALLEL_API_KEY="your-key"

Then start Harness:

npx --yes @deepseek-ai/[email protected] web

Open http://127.0.0.1:3080 in your browser. The first time you open Harness:

1. Open Settings → Models.
2. Add your DeepSeek API key.
3. Choose a workspace.
4. Start a new session.

4. Try a web search

You do not need to call a special Parallel tool. Ask Harness something that requires current information, for example:

Search the web for today's most important AI news and include links to your sources.

When Harness calls its built-in web_search tool, this plugin sends the search through Parallel.

Check that it worked

Inspect the profile:

npx --yes @deepseek-ai/[email protected] \
  --profile web --dump-config | \
  grep -E 'searchProvider: parallel|web-search-parallel'

You should see searchProvider: parallel and a web-search-parallel entry. Seeing the built-in web-search-deepseek plugin as well is normal—the searchProvider setting decides which provider handles searches.

For direct request evidence, start Harness with information-level Parallel logging:

PARALLEL_LOG=info npx --yes @deepseek-ai/[email protected] web

After a web_search call, the terminal should show a successful request to https://api.parallel.ai/v1/search. Review logs before sharing them.

Optional settings

The defaults work without extra configuration. If you want to tune the search, edit ~/.dsh/profiles/web/cordis.patch.yml:

- id: web-search-parallel
  config:
    mode: turbo
    maxCharsTotal: 12000
    maxCharsPerResult: 2000

Setting	Default	What it controls
mode	Parallel's advanced default	turbo, basic, or advanced search
maxCharsTotal	25000	Total excerpt characters returned to Harness
maxCharsPerResult	no limit	Excerpt characters kept for each result

Keep PARALLEL_API_KEY in the environment rather than putting it in this file, which is stored as readable text.

The plugin always sends requests to https://api.parallel.ai and ignores PARALLEL_BASE_URL.

If something goes wrong

• pnpm is not found: run npm install --global pnpm@10.
• Port 3080 is already in use: stop the older Harness process, then start Harness again.
• Parallel Search is unavailable: confirm PARALLEL_API_KEY is set in the same terminal that starts Harness.

Remove

npx --yes @deepseek-ai/[email protected] \
  plugin --profile web remove @parallel-web/dsh-web-search

Restart Harness after removing the plugin.

Development and security

See CONTRIBUTING.md for local development, tests, and the packed-artifact review workflow. Report vulnerabilities according to SECURITY.md.

For help, contact [email protected].

MIT licensed.