AI Agent Hub
Back to plugins
🧰

http-request-dsh-plugin

Web Tools Updated 2026.08.18

Run the following command in DeepSeek Harness:

dsh plugin install gao-gao-zai/http-request-dsh-plugin

Paste the following prompt into your AI chat to install this plugin:

Run dsh plugin install gao-gao-zai/http-request-dsh-plugin in your terminal to install the plugin; source code is available at https://github.com/gao-gao-zai/http-request-dsh-plugin

About this plugin

Making HTTP requests inside DeepSeek Harness used to mean spinning up an MCP Server or writing ad-hoc scripts. http-request-dsh-plugin registers the http_request tool as a pure DSH Plugin: no external process, no MCP stdio channel, no extra runtime to babysit. Once installed, outbound HTTP(S) calls become a native Harness capability.

The tool covers the full method set (GET, POST, PUT, PATCH, DELETE, HEAD, OPTIONS), accepts structured headers, query strings, JSON bodies, or plain text, and supports direct connections, system proxies, and manually specified HTTP(S) proxies. Response handling is equally thorough: configurable timeouts, character-level truncation for previews, full response-header inspection, and the ability to persist the entire response body to a file. On the security side, a credential-whitelist mechanism guards secret access end to end. On Windows the plugin reads from the Credential Manager; on Linux it shells out to pass and takes the first line of output. A request only triggers credential lookup when its target prefix matches the credentialPrefixes list; anything else receives a clean JSON error without ever touching the secret store.

This plugin is aimed at engineers and teams working inside DeepSeek Harness who routinely need data scraping, API debugging, automation glue, or backend integration—and who want those network calls to live inside the Harness process with a clear, auditable credential-security model, rather than maintaining a separate MCP service.

Use Cases

  • Calling third-party APIs directly inside Harness for data fetching and endpoint debugging
  • Injecting API keys or tokens securely without hard-coding secrets into config files
  • Firing authenticated HTTP(S) requests in automation pipelines and persisting the full response body to a file

Best For

  • Engineers debugging APIs or scraping data inside DeepSeek Harness
  • Development teams preferring in-process native tooling over maintaining separate MCP services
  • Developers who need whitelist-gated access to system credential stores for security compliance