AI Agent Hub
Back to plugins
🤖

dsh-subscriptions

Model Inference Updated 2026.08.26

Run the following command in DeepSeek Harness:

dsh plugin install GooDAnDReaDY/dsh-subscriptions

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

Run dsh plugin install GooDAnDReaDY/dsh-subscriptions in your DeepSeek Harness terminal to install this plugin, with the source available at https://github.com/GooDAnDReaDY/dsh-subscriptions

About this plugin

Running everyday agent and coding tasks on pay-as-you-go API credits adds up fast, while the ChatGPT Plus, Claude Pro, or xAI Premium subscriptions you already pay for sit idle in separate web interfaces. dsh-subscriptions bridges those paid subscriptions into DeepSeek Harness as first-class LLM providers: authenticate once via standard OAuth PKCE, then call them inside Harness exactly like any native model, eliminating per-token billing for daily work.

Reliability is a core design goal. Multi-account rotation pools automatically fail over to the next healthy account on HTTP 429, quota exhaustion, or rate-limit cooldowns, with preemptive quota switching and dynamic cooldown parsing from upstream headers. Every account slot accepts its own HTTP or SOCKS proxy with a one-click latency check. When all remote accounts are exhausted or unreachable, conversations seamlessly fall back to a local Ollama model, keeping offline demos and emergency sessions alive.

Security stays tight: OAuth tokens live only in the host encrypted credentials file and are never leaked through HTTP endpoints or the web UI; masked labels and quota bars are all the user sees. Headless and SSH environments get Device Code login plus a local loopback callback fallback. A privacy-mask toggle redacts emails across the entire interface, and the anonymized diagnostics report excludes tokens, credentials, and proxy URLs by design. Ideal for DeepSeek Harness builders who hold multiple AI subscriptions and want to drive everyday inference cost down to nearly zero.

Use Cases

  • Bridge existing ChatGPT, Claude, or xAI subscriptions into Harness to replace per-token API billing for daily agent tasks
  • Auto-rotate across a pool of accounts on 429 rate limits or quota exhaustion to keep workflows uninterrupted
  • Complete OAuth login via Device Code on headless SSH servers and invoke models without a browser

Best For

  • Developers building agent workflows on DeepSeek Harness who want to cut inference costs
  • Individual developers holding multiple paid AI subscriptions and aiming to drive everyday coding and reasoning overhead near zero
  • Technical users who need to invoke external AI models from headless servers, remote SSH sessions, or network-restricted environments