AI Agent Hub
Back to plugins
🖥️

dsh-plugin-ollama-usage

Client Updated 2026.09.16

Run the following command in DeepSeek Harness:

dsh plugin install CJ-SH/dsh-plugin-ollama-usage

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

Run dsh plugin install CJ-SH/dsh-plugin-ollama-usage in DeepSeek Harness to install the plugin from its source at https://github.com/CJ-SH/dsh-plugin-ollama-usage.

About this plugin

You are running Ollama Cloud as a backend inside dsh's web chat. Quota headroom, weekly burn rate—these details live on a separate Ollama dashboard you have to open in another tab. dsh-plugin-ollama-usage pulls the single GET /api/usage endpoint into the chat page itself: a compact panel sits below the composer in active sessions and in the hero area of new conversations, showing both the 5-hour rolling window and the weekly window with their respective percentages, refreshed every five minutes. When credentials are missing or the endpoint returns anything other than 2xx, the panel simply does not render—no placeholder, no error.

Credentials are handled in two modes. The default credential mode reuses your existing OLLAMA_API_KEY without the plugin ever writing to a shared secret; the key mode uses a plugin-specific reference OLLAMA_USAGE_API_KEY that the plugin owns exclusively. Keys are read only on the Host side, never shipped to the browser, and never appear in error messages. The plugin reads one endpoint, modifies no provider configuration, initiates no conversations, and its registration inserts only its own line with an independent route—zero coupling to other bundles.

If you already use Ollama Cloud inside dsh on a daily basis and do not want to context-switch to a dashboard every time you finish a prompt, this is the quiet little widget that answers the question for you. It tucks itself under the input box, stays out of the way when you are fine, and makes the limit visible the moment it matters.

Use Cases

  • See the 5-hour rolling Ollama Cloud quota percentage right under the composer without opening another tab
  • Get a visible burn-rate reminder in the hero area of new conversations before the window runs out
  • Switch between credential mode and key mode, or adjust baseURL in the plugin configuration card

Best For

  • Solo developers who use Ollama Cloud as their dsh web inference backend day in and day out
  • Technical users who want real-time quota visibility to avoid surprise 429 rate limits
  • Plugin users who prefer zero-intrusion, read-only endpoint integration with no provider config changes