agentsclimarketplace

Codex cli configuration

Skill Clarit-AI/Plexium/.claude/skills/codex-cli-configuration

This skill should be used when the user asks to "configure Codex CLI", "edit config.toml", "set up Codex settings", "change the Codex model", "configure approval policy", "set sandbox mode", "add an MCP server to Codex", "create a Codex profile", "configure model providers", "set up OTel for Codex", or mentions Codex CLI configuration, config.toml, or settings. Provides comprehensive configuration reference for OpenAI Codex CLI.From its SKILL.md

Install
npx -y skills add Clarit-AI/Plexium --skill codex-cli-configuration

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 5 stars5 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

4.2 KB, as published. Nobody here has run it

Codex CLI Configuration

Configure the OpenAI Codex CLI agent through layered TOML configuration files.

Config File Locations

FileScopePath
User configPersonal defaults~/.codex/config.toml
Project configPer-repo overrides.codex/config.toml
System configMachine-wide (Unix)/etc/codex/config.toml
RequirementsAdmin-enforcedrequirements.toml

Precedence (highest first)

  1. CLI flags and --config overrides
  2. Profile values (--profile <name>)
  3. Project config files (closest .codex/config.toml wins; trusted projects only)
  4. User config (~/.codex/config.toml)
  5. System config (/etc/codex/config.toml)
  6. Built-in defaults

Untrusted projects skip .codex/ layers entirely.

Quick Reference: Common Options

KeyValuesDefaultPurpose
modelstring (e.g. "gpt-5.4")built-inDefault model
approval_policyuntrusted, on-request, never, { granular = {...} }on-requestWhen to prompt for approval
sandbox_moderead-only, workspace-write, danger-full-accessread-onlyFilesystem/network sandbox level
web_searchdisabled, cached, livecachedWeb search mode
model_reasoning_effortminimal, low, medium, high, xhighmodel defaultReasoning depth
personalitynone, friendly, pragmaticmodel defaultCommunication style
model_providerprovider id from model_providers"openai"Which provider to use
file_openervscode, cursor, windsurf, nonevscodeCitation link scheme
profileprofile nameunsetDefault profile on startup

Feature Flags

Enable in [features] table or via codex --enable <name>:

FlagDefaultMaturityPurpose
codex_hooksfalseUnder developmentLifecycle hooks
multi_agenttrueStableSubagent collaboration
fast_modetrueStableFast mode + service_tier fast
shell_snapshottrueStableSnapshot shell env
undofalseStablePer-turn git undo
smart_approvalsfalseExperimentalGuardian reviewer subagent
appsfalseExperimentalChatGPT Apps/connectors

CLI One-off Overrides

# Dedicated flags
codex --model gpt-5.4

# Generic key/value (TOML syntax)
codex --config model='"gpt-5.4"'
codex --config sandbox_workspace_write.network_access=true
codex --config 'shell_environment_policy.include_only=["PATH","HOME"]'

Profiles

Define named presets in [profiles.<name>] and switch with codex --profile <name>:

[profiles.deep-review]
model = "gpt-5-pro"
model_reasoning_effort = "high"
approval_policy = "never"

Project Trust

[projects."/path/to/project"]
trust_level = "trusted"  # or "untrusted"

Additional Resources

Reference Files

  • references/config-options.md — Detailed common options with examples and values
  • references/advanced-config.md — Profiles, model providers, shell env policy, OTel, notifications, TUI, history, permissions
  • references/config-reference-table.md — Full searchable key/type/description reference
  • references/sample-config.md — Complete annotated config.toml with all sections

Example Configs

  • examples/minimal-config.toml — Minimal working config
  • examples/developer-config.toml — Common developer setup with profiles and providers

Scripts

  • scripts/validate-config.sh — Validate config.toml structure and common errors

External Source

For the latest configuration documentation: https://developers.openai.com/codex/config-basic

Keep looking

Skills are one crate of 610,220. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.