iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
If an MCP approval prompt never appears, first identify which runtime is making the tool call. In the Responses API, approval is controlled by the MCP tool’s require_approval policy and its filters; a policy of never or a filter that does not match the called tool can explain the missing pause. Codex plugin settings are a separate configuration surface, and Responses API or Agents SDK applications do not automatically inherit Codex Auto-review.
Start by identifying the runtime
“MCP approval” can refer to different controls. The Responses API exposes an approval policy on its MCP tool configuration. A Codex plugin can also specify approval modes for its MCP server and tools. Codex CLI or IDE behavior is another case: the sources cited here do not establish the exact setting precedence for a particular CLI or IDE version.
- Responses API or Agents SDK application: Check the API tool policy and the application’s own review flow.
- Codex plugin MCP server: Inspect the plugin’s server default and per-tool approval settings.
- Codex CLI or IDE MCP configuration: Gather the product version and relevant configuration scopes before assuming an API setting applies.
These controls should not be treated as interchangeable. OpenAI’s MCP documentation describes Responses API approval behavior, while its human-review guidance says, “Responses API and Agents SDK applications don’t automatically inherit Codex Auto-review.” OpenAI’s human-review guidance
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Check the Responses API approval policy
OpenAI’s Responses API MCP guide says MCP calls can be allowed automatically or restricted to require explicit developer approval. For the Responses API MCP tool, the require_approval setting can be always, never, or an approval filter. A never policy means the covered tools do not require approval.
#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
The same guide notes that, by default, OpenAI requests approval before sharing data with a connector or remote MCP server. That default does not override an explicit policy in your configured tool: inspect the value actually sent with the request.
What the policy choices mean
| Policy | Scope | What to check |
|---|---|---|
always |
All covered tools | Confirm this is the policy on the MCP tool configuration used by the request. |
never |
All covered tools | Approval is skipped for those tools. |
| Approval filter | Tools selected by the filter | Check the filter’s exact criteria against the tool that was called. |
The API guide also documents an approval-request item when a call requires approval, along with ways to skip approval for all tools or selected tools. If no request appears, determine whether the call was covered by an approval requirement at all before troubleshooting how the application displays or handles a request.
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
Compare the called tool with the filter
An approval filter can use tool names or read-only status. For a read-only filter, the API reference ties matching to the MCP server’s readOnlyHint annotation. A filter that requires approval only for certain names or for tools not marked read-only will not necessarily pause on every call.
The Tool Desk
Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →- Find the exact MCP tool name in the request and compare it with any name-based approval criteria.
- If the filter uses
read_only, check whether the server annotates the tool withreadOnlyHint. - Check both approval-required and approval-exempt criteria in the filter; a tool may fall into an exempt category.
- Confirm that the configured policy belongs to the MCP server and request that actually produced the call.
The Responses API reference describes the available approval-filter conditions. Do not infer that a tool is read-only from its name or apparent purpose; the documented filter depends on the server annotation.
Rank #3
- MAX II EPM240 CPLD Development Board Experiment Board Learning Breadboard
Add human review in an API application when needed
For a workflow built with the Responses API or Agents SDK, Codex Auto-review is not an automatic safety net. OpenAI states that those applications do not inherit it. If a person must approve an action, implement that review in the application’s harness and enforce it at the boundary where the consequential action occurs, rather than assuming a Codex interface setting will pause an API-driven workflow.
Keep the review boundary explicit: an API MCP approval request is a runtime control around a tool call, while application-level human review is logic your application must supply. Ensure the application does not proceed with the side effect until its required review has been completed.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Inspect plugin-scoped MCP settings separately
Plugin-scoped MCP configuration includes default_tools_approval_mode for the server and approval_mode for individual tools. Review both values in the plugin configuration that is active for the server. The available documentation establishes those settings exist, but does not establish how they interact with other Codex settings or which setting takes precedence in every version and configuration context. Do not assume the Responses API’s require_approval rules settle that question.
Recommended Free Tools
For plugin details, see OpenAI’s Codex plugin documentation. If the expected prompt is missing in a plugin workflow, verify the actual runtime and configuration before changing settings based on a different MCP integration.
For Codex CLI or IDE, collect configuration context first
The exact Codex CLI or IDE setting and precedence cannot be determined from the documented API behavior alone. Before making a setting-level diagnosis, record the Codex runtime and version, then identify which user, project, managed, or plugin configuration applies to the MCP server. Without that context, treating require_approval as the CLI or IDE approval switch risks changing an unrelated control.
Quick Recap
- Record whether the tool call originates in the Responses API, an SDK application, a plugin, the CLI, or an IDE.
- Capture the relevant runtime version and active MCP server configuration.
- For a plugin, inspect the server default and the individual tool’s approval mode.
- For an API workflow, inspect the actual
require_approvalvalue and filter rather than relying on Codex settings.
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.
