Bundle
dsh-plugin-openai-api
OpenAI-compatible /v1/chat/completions (streaming and non-streaming) and /v1/models endpoint on the DSH web server: one DSH agent session per (API key, X-Agent-Preset) pair.
- Source
- paulalesius
- License
- MIT
- Updated
- Updated 6 days ago
Readme
# dsh-plugin-openai-api
<p align="center"><img src="./banner.jpg" alt="dsh-plugin-openai-api banner"/></p>
An OpenAI-compatible endpoint on a running `dsh web` server:
```
GET /v1/models
POST /v1/chat/completions (non-streaming and SSE streaming)
```
## Installation
Install the plugin:
```bash
dsh plugin --profile web add /path/to/openai-api
```
(From a DSH source checkout, the same command is `pnpm dsh plugin …`.
A plain path links the source, so edits are picked up on restart. Fresh
checkout of this repo: recreate `node_modules` first —
`ln -s /path/to/deepseek-harness/apps/cli/node_modules node_modules`.)
Uninstall the plugin:
```bash
dsh plugin --profile web remove dsh-plugin-openai-api
```
Restart `dsh web` after either.
## Configuration
Append to `~/.dsh/profiles/web/cordis.patch.yml` (the install ships
`disabled: true`):
```yaml
- id: openai-api
disabled: false
config:
model: default
apiKeys:
- local-key
```
The `config` block replaces the plugin's defaults wholesale. Set
`disabled: true` to switch the endpoint off. Restart `dsh web` after
changes. All keys are optional:
| key | default | what it does |
| --- | --- | --- |
| `model` | `default` | The model id advertised at `/v1/models`. A request's `model` (absent, or this id) → the host's default model selection; any other id is passed through to the session. |
| `provider` | host default | The provider route created sessions use. |
| `apiKeys` | unset (no auth) | Non-empty array of client keys. When set, a request must present one of them (`Authorization: Bearer <key>`), else 401. A key identifies the caller; it does not select the preset. |
| `cwd` | the `dsh web` process's working directory | Working directory for created sessions. Must start with `/`. |
## Using the API
The contract is OpenAI's — it just works.
```js
import OpenAI from 'openai'
const client = new OpenAI({
baseURL: 'http://127.0.0.1:3080/v1',
apiKey: 'local-key', // one of the config's apiKeys
})
// The agent preset rides a header; omit it for the default preset.
const preset = { 'X-Agent-Preset': 'custom-agent-preset' }
const reply = await client.chat.completions.create(
{
model: 'default',
messages: [
{ role: 'user', content: 'What are you? ' },
{ role: 'assistant', content: "I'm a DSH agent." },
{ role: 'user', content: 'Write a haiku about that.' },
],
},
{ headers: preset },
)
console.log(reply.choices[0].message.content)
```
The same, without an SDK (the `Bearer` header only when you set
`apiKeys`):
```sh
curl -s http://127.0.0.1:3080/v1/chat/completions \
-H 'content-type: application/json' -H 'Authorization: Bearer local-key' \
-H 'X-Agent-Preset: custom-agent-preset' \
-d '{"messages":[{"role":"user","content":"hello"}]}'
```
## Headers
### `X-Agent-Preset`
The agent preset id the session joins — any id in
`~/.dsh/.agent-presets/`, plus the shipped presets.
- Absent → the default preset (`standard`).
- Unknown id → 400 with the roster; repeated or comma-joined → 400.
- Resolved at session creation only: the same (key, preset) pair always
resumes the same session; a different value is a different session.
Install
dsh plugin --profile web add github:paulalesius/dsh-openai-api#e4e47559248bd051d0bd7f86010c6633dc17f1af
Profile: web
With the hub plugin installed, ask your agent to install it by name — it resolves the same plan shown here.
dsh plugin --profile web add github:stvlynn/dsh.fish#path:packages/dsh-plugin-hub
install dsh-plugin-openai-api from the hub