Instantly MCP Server

Instructions

Implementation Reference

• src/instantly_mcp/tools/campaigns.py:302-344 (handler)

  The primary handler function implementing the list_campaigns tool logic. Fetches campaigns from the Instantly.ai API /campaigns endpoint, handles input parameters for pagination (starting_after, limit), search, and tags. Adds helpful pagination hints to the response.

  async def list_campaigns(params: Optional[ListCampaignsInput] = None) -> str:
      """
      List campaigns with cursor-based pagination (100 per page).
      
      PAGINATION: If response contains pagination.next_starting_after, there are 
      MORE results. Call again with starting_after=<that value> to get next page.
      Continue until pagination.next_starting_after is null.
      
      Note: search filters by campaign NAME only, not by status.
      To filter by status, use campaign_status in get_daily_campaign_analytics.
      
      Returns campaign list with status, lead counts, and performance metrics.
      """
      client = get_client()
      
      # Handle case where params is None (for OpenAI/non-Claude clients)
      # Set default limit=100 to return more results by default
      if params is None:
          params = ListCampaignsInput(limit=100)
      
      query_params = {}
      if params.limit:
          query_params["limit"] = params.limit
      else:
          # Default to 100 results if no limit specified
          query_params["limit"] = 100
      if params.starting_after:
          query_params["starting_after"] = params.starting_after
      if params.search:
          query_params["search"] = params.search
      if params.tag_ids:
          query_params["tag_ids"] = params.tag_ids
      
      result = await client.get("/campaigns", params=query_params)
      
      # Add pagination guidance for LLMs
      if isinstance(result, dict):
          pagination = result.get("pagination", {})
          next_cursor = pagination.get("next_starting_after")
          if next_cursor:
              result["_pagination_hint"] = f"MORE RESULTS AVAILABLE. Call list_campaigns with starting_after='{next_cursor}' to get next page."
      
      return json.dumps(result, indent=2)

• src/instantly_mcp/models/campaigns.py:86-108 (schema)

  Pydantic BaseModel defining the input schema (parameters) for the list_campaigns tool, including optional fields for limit, pagination cursor (starting_after), search term, and tag filtering.

  class ListCampaignsInput(BaseModel):
      """Input for listing campaigns with pagination."""
      
      # Use extra="ignore" to be tolerant of unexpected fields from LLMs
      model_config = ConfigDict(str_strip_whitespace=True, extra="ignore")
      
      limit: Optional[int] = Field(
          default=100, ge=1, le=100,
          description="Results per page (1-100, default: 100)"
      )
      starting_after: Optional[str] = Field(
          default=None,
          description="Pagination cursor - use value from pagination.next_starting_after to get next page"
      )
      search: Optional[str] = Field(
          default=None,
          description="Search by campaign NAME only (not status)"
      )
      tag_ids: Optional[str] = Field(
          default=None,
          description="Comma-separated tag IDs"
      )

• src/instantly_mcp/server.py:78-78 (registration)

  MCP tool registration annotation in the TOOL_ANNOTATIONS dictionary, marking list_campaigns as read-only (does not modify data). The tool is dynamically registered via the register_tools() function loop.

  "list_campaigns": {"readOnlyHint": True},