fix(provider): parse URL-based provider IDs

[obadakhalili]

Issue for this PR

Closes #39926

Type of change

• Bug fix
• New feature
• Refactor / code improvement
• Documentation

What does this PR do?

parseModel does not account for models connected through a well-known gateway. It expects providerID/modelID, but fails when the provider ID is itself a URL. Normal prompts work because they send the provider and model IDs separately, while commands combine them into a single string that the server must parse.

the PR fixes parseModel to account for gateway-connected models.

How did you verify your code works?

• create a mock gateway server to connect to it:

const host = "127.0.0.1"
const port = 8787
const origin = `http://${host}:${port}`
const model = process.env.ANTHROPIC_MODEL ?? "claude-sonnet-4-5"

const server = Bun.serve({
  hostname: host,
  port,
  async fetch(request) {
    const url = new URL(request.url)

    if (url.pathname === "/.well-known/opencode") {
      return Response.json({
        auth: {
          command: ["sh", "-c", 'printf %s "$ANTHROPIC_API_KEY"'],
          env: "ANTHROPIC_API_KEY",
        },
        config: {
          provider: {
            [origin]: {
              name: "Local Anthropic Gateway",
              npm: "@ai-sdk/anthropic",
              options: {
                apiKey: "{env:ANTHROPIC_API_KEY}",
                baseURL: `${origin}/v1`,
              },
              models: {
                [model]: {
                  name: model,
                  attachment: true,
                  reasoning: true,
                  tool_call: true,
                  limit: {
                    context: 200000,
                    output: 64000,
                  },
                  modalities: {
                    input: ["text", "image", "pdf"],
                    output: ["text"],
                  },
                },
              },
            },
          },
        },
      })
    }

    if (!url.pathname.startsWith("/v1/")) return new Response("Not found", { status: 404 })

    const headers = new Headers(request.headers)
    headers.delete("host")
    headers.delete("content-length")
    headers.delete("accept-encoding")

    const response = await fetch(`https://api.anthropic.com${url.pathname}${url.search}`, {
      method: request.method,
      headers,
      body: request.body,
      redirect: "manual",
    })
    const responseHeaders = new Headers(response.headers)
    responseHeaders.delete("content-encoding")
    responseHeaders.delete("content-length")
    return new Response(response.body, {
      status: response.status,
      statusText: response.statusText,
      headers: responseHeaders,
    })
  },
})

console.log(`Local Anthropic well-known gateway: ${server.url}`)
console.log(`Model: ${model}`)

• register the gateway:

export ANTHROPIC_API_KEY="your-key"
bun dev auth login http://127.0.0.1:8787

• start opencode with a test command:

export OPENCODE_CONFIG_CONTENT='{"command":{"gateway-test":{"template":"Reply with exactly COMMAND_OK"}}}'
bun dev

• select the gateway model in /models

• sending a prompt works

• before fix: /gateway-test fails with "Model not found"

• after fix: /gateway-test succeeds

Checklist

• I have tested my changes locally
• I have not included unrelated changes in this PR

[github-actions]

Thanks for updating your PR! It now meets our contributing guidelines. 👍

[simcop2387]

Can confirm as the original reporter that this does seem to fix the issue for me. Still trying to learn how to properly build and install from the branch to do more extensive testing but with npm run dev from within at least it's doing the right things.