An MCP connection does not universally provide 11 tools. It connects an agent to a server that publishes tool definitions; the tools an agent can use depend on that server, what the API successfully discovers, and any filters applied. The phrase “23 Actors” is not defined in the available OpenAI documentation, so those two counts cannot be reconciled without identifying the product, runtime, and server behind them.
What an MCP tool list actually represents
Model Context Protocol (MCP) lets an agent connect to a server that publishes tool definitions. OpenAI’s MCP connections guide describes connecting an MCP server to the Agents API. The MCP servers guide explains that the Responses API can discover the server’s tools and call them.
As an Amazon Associate I earn from qualifying purchases.
So a displayed list is best understood as tools available through a particular configured connection—not a universal set created by MCP itself. A server may publish tools, and API settings may further restrict which discovered tools are accessible.
Why you might see 11 tools
The number 11 could describe a particular server’s published tools, the tools imported in a particular setup, or a filtered set. The reviewed OpenAI documentation does not establish 11 as a default count for MCP connections. Its example tool lists demonstrate configuration and discovery, not a global default.
#1 Best Overall
- ESP32-C5 Core Processor: Equipped with ESP32-C5-WROOM-1 module, it supports dual-band Wi-Fi 6 and provides strong math for IoT edge AI applications
- 2.8" Touchscreen Display:Built-in 2.8" TFT color touchscreen, plug and play, support intuitive touch interactive operation
- ESP-Claw AI Smart Body Framework: Built-in ESP-Claw Chat Programming AI Smart Body Framework that supports event driving, structured memory, MCP communication, and custom skill extensions
- Multi-model LLM Compatible: ESP-Claw supports OpenAI style and Anthropic API, native compatible with major language models such as GPT, Qwen, Claude and DeepSeek
- (Wide Interface) Compatible with Arduino (USB-C), TF card slot, UART, FPC-IO and other interfaces, and is fully compatible with Arduino development environments, allowing for quick prototyping development
In the Responses API, allowed_tools can filter the tools the model may use. If your observed list is smaller than the server’s published list, check that setting and any other filtering in your application. Compare the server’s own tool list with the tools discovered by the API before treating the displayed count as the server’s full inventory.
What “23 Actors” could mean—and what is unknown
The available documentation does not define “Actors” or explain how Actors relate to MCP tools. It therefore cannot establish whether your 23 Actors are tools, agent components, objects in another product, or something else. Their absence from an MCP tool list does not, on its own, show that discovery failed: the two labels may refer to different kinds of things.
Rank #2
- ESP32-C5 Core Processor: Equipped with ESP32-C5-WROOM-1U module, it supports dual-band Wi-Fi 6 and provides strong math for IoT edge AI applications. FCC ID: 2AC7Z-ESPC5WROOMU
- 2.8" Touchscreen Display:Built-in 2.8" TFT color touchscreen, plug and play, support intuitive touch interactive operation
- ESP-Claw AI Smart Body Framework: Built-in ESP-Claw Chat Programming AI Smart Body Framework that supports event driving, structured memory, MCP communication, and custom skill extensions
- Multi-model LLM Compatible: ESP-Claw supports OpenAI style and Anthropic API, native compatible with major language models such as GPT, Qwen, Claude and DeepSeek
- (Wide Interface) Compatible with Arduino (USB-C), TF card slot, UART, FPC-IO and other interfaces, and is fully compatible with Arduino development environments, allowing for quick prototyping development
To investigate the mismatch, first identify the product or runtime that displays Actors and the MCP server label used for the connection. Then check whether that product exposes Actors as MCP tools at all. Without those details, claiming that the 23 Actors should appear in the list—or diagnosing why they do not—would be speculation.
Check the connection and its configuration
The Agents API guide documents three connection patterns. Which one you use affects where the connection runs and whether the server must be reachable from that location.
Rank #3
- MAX II EPM240 CPLD Development Board Experiment Board Learning Breadboard
| Connection pattern | Where the connection originates | What to check |
|---|---|---|
| HTTP from OpenAI’s service | OpenAI’s service | The server must be reachable from that service. |
| HTTP from the execution environment | The execution environment | Confirm the server is reachable from the environment running the connection. |
| stdio | The execution environment | Confirm the environment is available to start and communicate with the server. |
For Codex, MCP settings are stored in config.toml. The Codex MCP documentation describes user and trusted-project configuration scopes and says ChatGPT desktop, Codex CLI, and the IDE extension share this configuration. Check which scope supplies the server entry; a different active scope can mean you are inspecting a different configuration than the one in use.
A practical diagnosis
- Identify the runtime and connection. Record which product or API is showing the 11 tools, the configured server label, and whether the connection uses service-origin HTTP, environment-origin HTTP, or stdio.
- Inspect the server’s published tools. Compare that list with the tools the API reports as discovered. The MCP server and API documentation distinguish server publication from API discovery and calling.
- Check filters. In a Responses API setup, inspect
allowed_toolsand compare its contents with the discovered list. - Verify the active configuration scope. In Codex, inspect the applicable user or trusted-project entry in
config.tomland confirm it is the server you intend to use. - Define “Actors” in the originating product. Find out whether those objects are exposed by that product as MCP tools. If they are expected to appear, compare their published tool definitions with the discovered list; if not, they may not belong in that list.
This comparison separates three questions that can otherwise be conflated: what the server publishes, what the API discovers, and what the runtime permits the agent to use. The documentation does not identify the particular server or Actor system behind the two counts in the title.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
