AI Agent Hub
Back to plugins
⚙️

dsh-proxy-env

Workflow Updated 2026.09.07

Run the following command in DeepSeek Harness:

dsh plugin install localSummer/dsh-proxy-env

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

Run dsh plugin install localSummer/dsh-proxy-env in your terminal to install this plugin; the full source is available at https://github.com/localSummer/dsh-proxy-env .

About this plugin

Making remote model API calls, Git operations, or npm installs inside DSH often means every subprocess needs its own proxy variables. Scattering HTTP_PROXY exports across shell scripts is tedious and easy to get wrong. dsh-proxy-env takes over the global dispatcher and child-process environment at the Host level, applying a single proxy policy across the entire DSH process tree in one deterministic switch.

Two granularities are available. The all mode writes HTTP(S)_PROXY, ALL_PROXY, and NO_PROXY for model calls, subprocesses, Git, npm, and more, while merging the machine-level no_proxy list and forcing localhost and 127.0.0.1 into the exclusion set. The include mode restricts proxying to a hand-picked host list for in-process fetches only, leaving child-process environments untouched. A built-in safety net falls back to a direct connection and raises an alert when the proxy is unreachable, so your toolchain is never hard-blocked. A dedicated Proxy Environment card on the Web settings page, the /proxy slash command, and the proxy_env tool let you toggle, inspect, or reconfigure the proxy at any time.

If you frequently call remote model endpoints or rely on external dev tooling inside DSH and your network demands selective proxying, this plugin turns a pile of ad-hoc shell variables into one globally effective, single-point configuration. It is ideal for developers who want a clean, safe, and auditable proxy layer without touching every subprocess individually.

Use Cases

  • Calling remote model APIs inside DSH and needing a single global proxy instead of per-process HTTP_PROXY exports
  • npm and Git subprocesses require different proxy policies, controlled by a hand-picked host list
  • A proxy endpoint goes unreachable and the workflow must fall back to direct connection without hard-blocking

Best For

  • AI application developers who make frequent remote model API calls within DSH
  • Backend engineers whose subprocess toolchains (Git, npm) rely on selective proxying
  • Teams operating under managed network egress that need per-host proxy policies