pei-lin-001/dsh-ollama-cloud
Adds Ollama Cloud as a DeepSeek Harness provider route from a single API key, with no local Ollama install and a sidebar quota chip.
On boot, on a credential change and on demand, the plugin pulls the account's model catalog — GET /v1/models for the ids, then one POST /api/show per model — and writes it as one llm-pi-ai route, so the Ollama Cloud models appear in the model picker without a local Ollama install. The plugin supplies only the catalog and the credential; request handling stays with the shared provider layer. A sidebar chip shows the remaining quota. Because everything is keyed off the credentials service, rotating the key updates the route without editing configuration files.
Install
dsh plugin --profile web add git:github.com/pei-lin-001/dsh-ollama-cloudCopied verbatim from the README's Install block, web-profile line; the same block lists the desktop-profile form and a development checkout that links $PWD before running ./scripts/link-peers.sh. The registry name dsh-ollama-cloud returns 404, so the documented git: spec is the install path. After installing, open Settings → Ollama Cloud, paste a key and save-and-sync.
Compatibility
Needs an Ollama Cloud API key from ollama.com/settings/keys. The key is written to the credentials service at $DSH_HOME/.credentials.yaml and never reaches the browser; an OLLAMA_API_KEY in the process environment also works but shadows a stored key, and in that state the credentials service refuses writes. The catalog is written to the llm-pi-ai settings section as route ollama-cloud, and chat, tools, vision and thinking are served by dsh-llm-pi-ai with the openai-completions API against https://ollama.com/v1.
Details
- Repo: pei-lin-001/dsh-ollama-cloud
- Category: Models & Providers
- Stars: 0
- Version: 0 stars, JavaScript, MIT, created 2026-10-18; install source is the GitHub repository; English README with a README.zh.md translation.
Alternatives
elves-ai/dsh-llm-nowcoding · penguin-oo/dsh-delegate-router · lcestou/dsh-oh-my-claude