All docs
2 min read

Install for Claude Code

1. Mint a token

Open API tokens, create a token, copy the value (shown once).

A safe starter set of abilities:

forms:read, submissions:read, submissions:write, webhooks:read, insights:read.

Add forms:write or webhooks:write only if the agent should mutate things.

2. Add the server

One-liner via the CLI:

claude mcp add formspring \
  --transport http \
  --url https://formspring.io/mcp-server \
  --header "Authorization: Bearer your-api-token"

Add --scope project to keep it in .mcp.json instead of ~/.claude.json.

Or edit the config file directly:

{
  "mcpServers": {
    "formspring": {
      "url": "https://formspring.io/mcp-server",
      "headers": {
        "Authorization": "Bearer your-api-token"
      }
    }
  }
}

3. Verify

/mcp

You should see formspring with its tools loaded.

Try a prompt

Paste any of these into a Claude Code session:

"List my forms."

"Triage the last 50 spam submissions on my contact form. Group them by reason and propose two new blocked-emails entries."

"Audit every webhook on every form. Tell me which are unhealthy and why."

More in Prompt examples →.

Slash-command shortcuts

Save reusable workflows to .claude/commands/triage-spam.md:

Triage today's spam folder for the contact form.

1. Call list_submissions with folder: "spam", limit: 50
2. Decide keep / mark not-spam / delete for each
3. Apply via bulk_submissions
4. Summarize what you did

Run with /triage-spam.

What's next