Connect AI agents to your feedback boards

UserPulse exposes a small, sharply-scoped HTTP API designed for LLM tool-use. Drop the tool definitions below into Claude or GPT, point at your UserPulse base URL, and your agent can triage, file, and close feedback.

The pattern

Issue a key in Settings → API keys.
Store it in USERPULSE_KEY.
Define tools that wrap the relevant endpoints.
Inject Authorization: Bearer ${USERPULSE_KEY} in your tool runner, never expose it to the model.

Claude tool definitions

[
  {
    "name": "list_feedback",
    "description": "List feedback items on a UserPulse board.",
    "input_schema": {
      "type": "object",
      "properties": {
        "slug": { "type": "string" },
        "status": { "type": "string", "enum": ["open","in-progress","completed","closed"] },
        "limit": { "type": "integer", "default": 50 }
      },
      "required": ["slug"]
    }
  },
  {
    "name": "create_feedback",
    "description": "File a new feedback item on a UserPulse board.",
    "input_schema": {
      "type": "object",
      "properties": {
        "slug": { "type": "string" },
        "title": { "type": "string" },
        "description": { "type": "string" }
      },
      "required": ["slug","title"]
    }
  },
  {
    "name": "update_feedback_status",
    "description": "Move a feedback item to a new status.",
    "input_schema": {
      "type": "object",
      "properties": {
        "id": { "type": "string" },
        "status": { "type": "string", "enum": ["open","in-progress","completed","closed"] }
      },
      "required": ["id","status"]
    }
  }
]

OpenAI function calling

[
  {
    "type": "function",
    "function": {
      "name": "list_feedback",
      "description": "List feedback items on a UserPulse board.",
      "parameters": {
        "type": "object",
        "properties": {
          "slug": { "type": "string" },
          "status": { "type": "string", "enum": ["open","in-progress","completed","closed"] }
        },
        "required": ["slug"]
      }
    }
  }
]

MCP servers

Building a Model Context Protocol server? Wrap each endpoint as an MCP tool. The OpenAPI spec at /api/openapi.json can be consumed by tools like @modelcontextprotocol/server-openapi to generate one automatically.

Safety

Use read-only keys for triage agents that should never write.
Use board-scoped keys when the agent only owns one product surface.
Revoke a key the moment it leaks — agents stop working immediately.