ai-sdk-provider-dsh

AI SDK provider that drives a DeepSeek Harness (dsh) runtime as a language model.

dsh is a full agent harness (agent loop, tools, skills, MCP, sessions) — this provider wraps a dsh runtime subprocess behind the AI SDK LanguageModel interface, so you can drive a harness agent from AI SDK generateText / streamText the same way ai-sdk-provider-claude-code drives Claude Code.

Compatibility

Implements the LanguageModelV3 specification (specificationVersion: 'v3'), which is the shared interface across AI SDK majors:

AI SDK	@ai-sdk/provider	Status
ai@^6	@ai-sdk/provider@^3	✅ supported
ai@^7	@ai-sdk/provider@^4	✅ supported (V3 models are first-class in v7)

• Node.js >=22.19
• ESM only
• Pins the DeepSeek Harness SDK family to 0.1.0-rc.6

Install

npm install ai-sdk-provider-dsh

The dsh runtime is bundled: the package ships a default runtime composition (runtime/cordis.yml) and the dsh-jsonrpc-agent bin (via @deepseek-ai/dsh-sdk-jsonrpc-demo), so a provider instance spawns a working runtime out of the box. All runtime plugins are pinned exact versions in dependencies.

Usage

import { generateText, streamText } from "ai";
import { createDsh } from "ai-sdk-provider-dsh";

const dsh = createDsh({
  runtime: {
    provider: "deepseek-official",
    model: "deepseek-v4-flash",
    // credentials for the runtime subprocess:
    env: {
      DEEPSEEK_API_KEY: process.env.DEEPSEEK_API_KEY,
      // DEEPSEEK_BASE_URL: optional OpenAI-compatible gateway override
    },
  },
});

// streamText (v6: `system` / v7: `instructions`)
const result = streamText({
  model: dsh.languageModel("deepseek-v4-flash"),
  system: "You are a coding agent.",
  prompt: "run the tests",
});

Runtime options

Option	Default	Meaning
provider / model	required	model route passed to the runtime handshake
env	inherits process.env	env for the runtime subprocess (credentials, DSH_CWD, DSH_SESSION_ROOT, …)
cwd	process.cwd()	subprocess working directory
configPath	bundled runtime/cordis.yml	a different cordis.yml composition
binPath	bundled dsh-jsonrpc-agent	a different runtime bin
command / args	node + [bin, config]	full custom launch vector (both required if overridden)
maxTokens	—	positive output-token cap per root-agent request
sessionId	fresh UUID	fixed session id; keep it to continue one harness session across turns

The bundled default composition exposes bash (foreground), read/write/edit, subagent, and todo tools. For environments that cannot build node-pty (no Linux prebuild), use the no-pty minimal composition shipped as runtime/cordis.minimal.yml:

runtime: {
  provider: "deepseek-official",
  model: "deepseek-v4-flash",
  configPath: require.resolve("ai-sdk-provider-dsh/runtime/cordis.minimal.yml"),
}

How it works

• Each provider instance spawns (lazily) one dsh runtime subprocess speaking stdio JSON-RPC (the dsh SDK protocol).
• doGenerate / doStream translate AI SDK LanguageModelV3CallOptions into a dsh prompt, then map the runtime's session.event stream back into AI SDK stream parts (text-delta, tool-input-*, tool-call, finish).
• Tool execution happens inside the harness (dsh owns the agent loop and tools), like Claude Code providers — the provider is a thin pass-through.
• There is no mid-turn cancel on the SDK wire: aborting a turn means closing the runtime (close()); the runtime's append-only session log survives restarts for follow-up turns.

License

MIT