helphow to set trusted level codex[projects."/Users/kian/Developer"]trust_level = "trusted" | OpenAI | Page 1

plain moth Feb 10, 2026, 9:09 PM

#

I have the above config set, but it still asks me to trust directories that are subdirectories of them every time I open codex.

open auroraBOT Feb 10, 2026, 9:09 PM

#

Hi @plain moth. I'm kapa.ai, your friendly support bot. I will try to answer your question by searching through the knowledge base. :wave:

#

The knowledge sources you provided don’t mention the projects."…".trust_level = "trusted" setting or how Codex decides whether a directory is “trusted,” so I can’t give a definitive, config-level fix.

What is covered is:

In the VS Code extension, you can reduce prompts by choosing the Agent (full access) / “full permissions” model, which avoids frequent approvals for actions inside the repo. [How to give permissions]
Codex behavior can also be affected by VS Code’s remote / dev container settings (e.g., remote.extensionKind), which can cause Codex to see a different path than you expect (host vs container), leading to unexpected prompts. [Dev container behavior]

Because there’s no explicit documentation in these sources about:

how trust_level is parsed,
whether it applies recursively to subdirectories,
or whether the path must match Codex’s effective working directory (e.g., inside a container),

I can’t reliably say why your specific projects."/Users/kian/Developer"]trust_level = "trusted" isn’t being honored.

Given the limits of the docs here, the most I can suggest (without guessing) is:

In the VS Code Codex extension, set the model / agent mode to the “full access” / “Agent (full access)” option to reduce per-action prompts. [How to give permissions]
If you’re using dev containers or remote setups, confirm Codex is actually running in the environment that sees the same path you’ve configured as trusted (e.g., not pinned to the host via remote.extensionKind). [Dev container behavior]

#

Beyond that, the knowledge sources don’t provide enough detail on the `trust_level` config itself, so for this specific issue you’d likely need to check the Codex CLI/IDE docs directly or contact OpenAI support.

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:

#

@plain moth Kapa.ai is still learning and improving, please let me know how I did by reacting below.

plain moth Feb 10, 2026, 9:12 PM

#

no luck

You are running Codex in /Users/kian/Developer/regex

Since this folder is version controlled, you may wish to allow Codex to work in this folder without asking for approval.

› 1. Yes, allow Codex to work in this folder without asking for approval
2. No, ask me to approve edits and commands

plain moth Feb 10, 2026, 11:36 PM

#

@open aurora
does Codex CLI support codex -p "<prompt>" --model codex-5.3

open auroraBOT Feb 10, 2026, 11:36 PM

#

Let me check the knowledge base for you again. One moment! :books:

#

The knowledge sources don’t list any codex-5.3 model name, nor do they document a codex -p "<prompt>" --model codex-5.3 invocation syntax for the Codex CLI, so I can’t confirm that this exact command is supported.

From the docs we do have:

Current Codex-tuned API models are named like gpt-5.2-codex and gpt-5.1-codex-max, and these are the recommended models for Codex / coding workflows. [Codex prompting; Code generation guide; GPT‑5.2 FAQ]
The Codex CLI and SDK use these Codex model names (e.g., gpt-5.2-codex, gpt-5.1-codex-max) rather than a codex-5.3 style name. [Code generation guide; Model release notes]

Because there is no mention of a `codex-5.3` model or that specific CLI flag combination in the provided sources, I can’t say that `codex -p "<prompt>" --model codex-5.3` is valid. Based on the docs, you should instead select one of the documented Codex models (for example, `gpt-5.2-codex` or `gpt-5.1-codex-max`) in your Codex CLI configuration or command.

#helphow to set trusted level codex[projects."/Users/kian/Developer"]trust_level = "trusted"

Beyond that, the knowledge sources don’t provide enough detail on the trust_level config itself, so for this specific issue you’d likely need to check the Codex CLI/IDE docs directly or contact OpenAI support.

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:

Beyond that, the knowledge sources don’t provide enough detail on the `trust_level` config itself, so for this specific issue you’d likely need to check the Codex CLI/IDE docs directly or contact OpenAI support.