AI Content Drop in VS Code

VS Code reads MCP servers from an mcp.json file — one at .vscode/mcp.json inside a workspace for just that repo, or the user-level one that the MCP: Open User Configuration command opens for every workspace. Browsing works immediately with no credential; generating needs an API key.

Install with one click

The install button on the VS Code tab of the MCP page opens VS Code through its vscode:mcp/install URL handler with one server configuration already filled in: the name aicontentdrop, the type http, and the URL https://aicontentdrop.com/mcp. VS Code shows the configuration, asks you to confirm, and writes the entry for you.

The handler takes one server configuration and no secret, so the API key is a separate step below.

Add it by hand

If the link does nothing in your build, save this as .vscode/mcp.json in the workspace:


{
  "servers": {
    "aicontentdrop": {
      "type": "http",
      "url": "https://aicontentdrop.com/mcp"
    }
  }
}

VS Code shows a Start action above the entry. Start it, then open the Chat view in agent mode and check the tools picker: nine tools should appear under aicontentdrop.

Verify without a credential

Ask the agent:

> Using AI Content Drop, list the video models under 20 credits.

That is list_models, which needs no key. Model ids and credit costs coming back means the connection works.

Authorize

Create a key in Settings → Integrations. Keys are shown once and look like acd_live_….

Do not paste the key into the file. Add an inputs entry instead, so VS Code asks for it once, stores it in its secret storage, and sends it as a header:


{
  "inputs": [
    {
      "type": "promptString",
      "id": "acd-api-key",
      "description": "AI Content Drop API key",
      "password": true
    }
  ],
  "servers": {
    "aicontentdrop": {
      "type": "http",
      "url": "https://aicontentdrop.com/mcp",
      "headers": {
        "Authorization": "Bearer ${input:acd-api-key}"
      }
    }
  }
}

Restart the server from the entry's action. VS Code prompts for the key on the first start and does not ask again. The five account tools (generate_video, generate_image, get_generation, list_generations, get_account) start working; the other four never needed the key.

First generation

> Quote a 5-second Kling 3.0 clip at 1080p, then generate one of a matcha tin > on a linen backdrop, and poll it until it is done.

The agent should call estimate_credit_cost, which returns a quote_id valid for 15 minutes, then generate_video with that quote_id, then get_generation until the job reports completed. Credits are charged on success only — a failed render costs nothing. A retry with the same quote_id returns the original job rather than a second charge.

Troubleshooting

The install link does nothing. The vscode: handler is registered by VS Code itself; if the browser cannot open it, use the hand-installed file above. It produces the same entry.

The server shows an error in the file. Check the JSON parses; a trailing comma is the usual cause. Then confirm the URL is https://aicontentdrop.com/mcp with no trailing slash and the type is http.

Account tools answer "requires authorization". That is the expected reply with no key, or with one that has been revoked. Re-check the header: it is Bearer followed by the input reference. If the stored key is stale, give the inputs entry a new id (and reference it in the header); VS Code prompts for the new id on the next start. The four public tools never need a key.

A generation returns a job id and nothing else. That is correct. Generation is submit-and-poll; ask the agent to call get_generation with the id.

QUOTE_REQUIRED, QUOTE_EXPIRED or QUOTE_MISMATCH. Every generation needs a fresh quote_id from estimate_credit_cost for the same model, duration and resolution. Quote again and retry.