{"id":2965,"date":"2026-04-25T18:50:15","date_gmt":"2026-04-25T18:50:15","guid":{"rendered":"https:\/\/aiopsschool.com\/blog\/?p=2965"},"modified":"2026-04-25T18:50:15","modified_gmt":"2026-04-25T18:50:15","slug":"complete-codex-slash-commands-and-cli-options-guide-updated-april-2026","status":"publish","type":"post","link":"https:\/\/aiopsschool.com\/blog\/complete-codex-slash-commands-and-cli-options-guide-updated-april-2026\/","title":{"rendered":"Complete Codex \/ Slash Commands and CLI Options Guide \u2014 Updated April 2026"},"content":{"rendered":"\n

OpenAI Codex has grown from a simple terminal coding assistant into a full workflow tool with terminal UI commands, desktop-app commands, IDE commands, cloud tasks, MCP support, plugins, custom prompts, skills, and automation-friendly CLI flags. This guide explains what appears when you type \/<\/code> inside Codex, what each command does, and which related codex<\/code> terminal options are worth knowing.<\/p>\n\n\n\n

1. What happens when you type \/<\/code> in Codex?<\/h2>\n\n\n\n

In the Codex CLI, typing \/<\/code> in the composer opens the slash-command popup. You can keep typing to filter commands. If Codex is already running a task, you can type a slash command and press Tab<\/code> to queue it for the next turn. (OpenAI Developers<\/a>)<\/p>\n\n\n\n

Slash commands are not the same as shell commands. They control the Codex session itself: model selection, approvals, review mode, conversation reset, session status, MCP tools, plugins, agent threads, and more.<\/p>\n\n\n\n

2. Master list: Codex CLI slash commands<\/h2>\n\n\n\n
Command<\/th>Purpose<\/th>Best use case<\/th><\/tr><\/thead>
\/permissions<\/code><\/td>Change what Codex can do without asking first.<\/td>Switch between safer and more autonomous modes mid-session.<\/td><\/tr>
\/sandbox-add-read-dir<\/code><\/td>Add sandbox read access to another directory.<\/td>Windows-only; useful when Codex needs to read an absolute path outside current roots.<\/td><\/tr>
\/agent<\/code><\/td>Switch active agent thread.<\/td>Inspect or continue work in a spawned subagent thread.<\/td><\/tr>
\/apps<\/code><\/td>Browse apps\/connectors and insert them into the prompt.<\/td>Attach an app as $app-slug<\/code> before asking Codex to use it.<\/td><\/tr>
\/plugins<\/code><\/td>Browse installed and discoverable plugins.<\/td>Inspect, install, or manage plugin tools.<\/td><\/tr>
\/clear<\/code><\/td>Clear terminal and start a fresh chat.<\/td>Reset the visible UI and conversation together.<\/td><\/tr>
\/compact<\/code><\/td>Summarize the visible conversation to free context.<\/td>Use during long sessions to preserve essentials while reducing context usage.<\/td><\/tr>
\/copy<\/code><\/td>Copy the latest completed Codex output.<\/td>Grab the latest finished answer or plan; Ctrl+O<\/code> also works.<\/td><\/tr>
\/diff<\/code><\/td>Show Git diff, including untracked files.<\/td>Review Codex\u2019s local edits before committing.<\/td><\/tr>
\/exit<\/code><\/td>Exit the CLI.<\/td>Same as \/quit<\/code>.<\/td><\/tr>
\/experimental<\/code><\/td>Toggle experimental features.<\/td>Enable optional workflows such as subagents where available.<\/td><\/tr>
\/feedback<\/code><\/td>Send feedback\/logs to Codex maintainers.<\/td>Report issues or diagnostics.<\/td><\/tr>
\/init<\/code><\/td>Generate an AGENTS.md<\/code> scaffold.<\/td>Create persistent repository instructions for Codex.<\/td><\/tr>
\/logout<\/code><\/td>Sign out of Codex.<\/td>Clear local credentials on a shared machine.<\/td><\/tr>
\/mcp<\/code><\/td>List configured MCP tools.<\/td>Check which external MCP tools Codex can call.<\/td><\/tr>
\/mention<\/code><\/td>Attach a file or folder to the conversation.<\/td>Point Codex at specific code or context.<\/td><\/tr>
\/model<\/code><\/td>Choose active model and reasoning effort where available.<\/td>Switch models before a task.<\/td><\/tr>
\/fast<\/code><\/td>Toggle\/check Fast mode for supported models.<\/td>Use when speed matters more than deeper reasoning.<\/td><\/tr>
\/plan<\/code><\/td>Switch to plan mode, optionally with a prompt.<\/td>Ask Codex for a plan before implementation.<\/td><\/tr>
\/personality<\/code><\/td>Choose response style.<\/td>Pick friendly<\/code>, pragmatic<\/code>, or none<\/code>.<\/td><\/tr>
\/ps<\/code><\/td>Show experimental background terminals and recent output.<\/td>Check long-running background work.<\/td><\/tr>
\/stop<\/code><\/td>Stop all background terminals.<\/td>Cancel background commands from the session.<\/td><\/tr>
\/fork<\/code><\/td>Fork the current conversation into a new thread.<\/td>Explore another approach without losing the original thread.<\/td><\/tr>
\/resume<\/code><\/td>Resume a saved conversation.<\/td>Continue a previous CLI session.<\/td><\/tr>
\/new<\/code><\/td>Start a new conversation in the same CLI session.<\/td>Reset chat context without quitting Codex.<\/td><\/tr>
\/quit<\/code><\/td>Exit the CLI.<\/td>Same as \/exit<\/code>.<\/td><\/tr>
\/review<\/code><\/td>Ask Codex to review the working tree.<\/td>Review local changes before commit or PR.<\/td><\/tr>
\/status<\/code><\/td>Show session configuration and token usage.<\/td>Confirm model, approvals, writable roots, and context capacity.<\/td><\/tr>
\/debug-config<\/code><\/td>Print config layer and requirements diagnostics.<\/td>Debug config precedence and policy requirements.<\/td><\/tr>
\/statusline<\/code><\/td>Configure TUI status-line fields.<\/td>Choose footer items such as model, context, git, tokens, limits, session.<\/td><\/tr>
\/title<\/code><\/td>Configure terminal window\/tab title fields.<\/td>Customize title fields such as project, branch, model, task progress.<\/td><\/tr>
\/theme<\/code><\/td>Preview and save a TUI theme.<\/td>Documented in the CLI features guide; may depend on version\/build.<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n

The official slash-command page also notes that \/approvals<\/code> still works as an alias, but it no longer appears in the slash popup list. (OpenAI Developers<\/a>) The CLI features page separately documents \/theme<\/code> for previewing and saving a preferred theme. (OpenAI Developers<\/a>)<\/p>\n\n\n\n

3. Commands you may see outside CLI: Codex app and IDE<\/h2>\n\n\n\n

The Codex desktop app has a smaller slash list, and OpenAI notes that available commands vary by environment and access. The documented app slash commands are \/feedback<\/code>, \/mcp<\/code>, \/plan-mode<\/code>, \/review<\/code>, and \/status<\/code>; enabled skills can also appear in the slash list, and skills can be explicitly invoked with $<\/code>. (OpenAI Developers<\/a>)<\/p>\n\n\n\n

The Codex IDE extension has its own slash-command set: \/auto-context<\/code>, \/cloud<\/code>, \/cloud-environment<\/code>, \/feedback<\/code>, \/local<\/code>, \/review<\/code>, and \/status<\/code>. These are aimed at switching between local\/cloud execution, selecting cloud environments, controlling automatic context, and reviewing code from inside the IDE. (OpenAI Developers<\/a>)<\/p>\n\n\n\n

4. Custom slash commands: \/prompts:name<\/code><\/h2>\n\n\n\n

Codex also supports custom prompt commands, but OpenAI now marks custom prompts as deprecated and recommends skills for reusable instructions. Existing custom prompts still let you create Markdown files under ~\/.codex\/prompts<\/code> and invoke them from the slash menu, for example \/prompts:draftpr<\/code>. Custom prompt files can include YAML front matter such as description<\/code> and argument-hint<\/code>, plus placeholders like $1<\/code>, $ARGUMENTS<\/code>, $FILE<\/code>, or $TICKET_ID<\/code>. (OpenAI Developers<\/a>)<\/p>\n\n\n\n

Example:<\/p>\n\n\n\n

mkdir -p ~\/.codex\/prompts\n<\/code><\/pre>\n\n\n\n
Create:<\/p>\n\n\n\n
~\/.codex\/prompts\/draftpr.md\n<\/code><\/pre>\n\n\n\n
Example command after restart:<\/p>\n\n\n\n
\/prompts:draftpr FILES=\"src\/app.ts src\/api.ts\" PR_TITLE=\"Add API loading state\"\n<\/code><\/pre>\n\n\n\n
Codex expands the prompt file, replaces placeholders, and sends the expanded instruction into the current session.<\/p>\n\n\n\n
5. Most useful slash-command workflows<\/h2>\n\n\n\n
Switch model before a difficult task<\/h3>\n\n\n\n
Use:<\/p>\n\n\n\n
\/model\n<\/code><\/pre>\n\n\n\n
Pick a stronger reasoning model for complex architecture, debugging, or refactoring. Then run:<\/p>\n\n\n\n
\/status\n<\/code><\/pre>\n\n\n\n
to verify the active model and session settings. (OpenAI Developers<\/a>)<\/p>\n\n\n\n
Ask for a plan before coding<\/h3>\n\n\n\n
Use:<\/p>\n\n\n\n
\/plan\n<\/code><\/pre>\n\n\n\n
Or inline:<\/p>\n\n\n\n
\/plan Propose a migration plan for this service\n<\/code><\/pre>\n\n\n\n
This is best before multi-file changes, migrations, refactors, or anything that may require review before implementation. (OpenAI Developers<\/a>)<\/p>\n\n\n\n
Review changes before committing<\/h3>\n\n\n\n
Use:<\/p>\n\n\n\n
\/diff\n<\/code><\/pre>\n\n\n\n
Then:<\/p>\n\n\n\n
\/review\n<\/code><\/pre>\n\n\n\n
\/diff<\/code> shows the local Git diff, while \/review<\/code> asks a dedicated Codex reviewer to inspect the working tree and return prioritized findings. (OpenAI Developers<\/a>)<\/p>\n\n\n\n
Recover context in long sessions<\/h3>\n\n\n\n
Use:<\/p>\n\n\n\n
\/compact\n<\/code><\/pre>\n\n\n\n
This summarizes the visible conversation to free context while preserving the key decisions and instructions. Use it before a long follow-up task rather than waiting until the session feels overloaded. (OpenAI Developers<\/a>)<\/p>\n\n\n\n
Start fresh without quitting<\/h3>\n\n\n\n
Use:<\/p>\n\n\n\n
\/clear\n<\/code><\/pre>\n\n\n\n
This is different from Ctrl+L<\/code>: \/clear<\/code> starts a new conversation, while Ctrl+L<\/code> only clears the terminal view. (OpenAI Developers<\/a>)<\/p>\n\n\n\n
6. Master list: important codex<\/code> CLI flags<\/h2>\n\n\n\n
The base codex<\/code> command accepts global flags, and most defaults come from ~\/.codex\/config.toml<\/code>. Command-line -c key=value<\/code> overrides take precedence for that invocation. (OpenAI Developers<\/a>)<\/p>\n\n\n\n
Flag<\/th>Meaning<\/th><\/tr><\/thead>
--add-dir path<\/code><\/td>Grant additional directories write access alongside the main workspace.<\/td><\/tr>
--ask-for-approval, -a untrusted|on-request|never<\/code><\/td>Control when Codex pauses for approval.<\/td><\/tr>
--cd, -C path<\/code><\/td>Set working directory before Codex starts.<\/td><\/tr>
--config, -c key=value<\/code><\/td>Override config values for this run.<\/td><\/tr>
--dangerously-bypass-approvals-and-sandbox<\/code>, --yolo<\/code><\/td>Run without approvals or sandboxing; use only in hardened environments.<\/td><\/tr>
--disable feature<\/code><\/td>Force-disable a feature flag.<\/td><\/tr>
--enable feature<\/code><\/td>Force-enable a feature flag.<\/td><\/tr>
--full-auto<\/code><\/td>Low-friction preset using on-request<\/code> approvals and workspace-write<\/code> sandbox.<\/td><\/tr>
--image, -i path[,path...]<\/code><\/td>Attach one or more image files to the initial prompt.<\/td><\/tr>
--model, -m string<\/code><\/td>Override configured model.<\/td><\/tr>
--no-alt-screen<\/code><\/td>Disable alternate-screen mode for the TUI.<\/td><\/tr>
--oss<\/code><\/td>Use the local open-source model provider; requires Ollama.<\/td><\/tr>
--profile, -p string<\/code><\/td>Load a named config profile.<\/td><\/tr>
--remote ws:\/\/host:port<\/code> or wss:\/\/host:port<\/code><\/td>Connect TUI to a remote app-server endpoint.<\/td><\/tr>
--remote-auth-token-env ENV_VAR<\/code><\/td>Send bearer token from an environment variable for remote mode.<\/td><\/tr>
--sandbox, -s read-only|workspace-write|danger-full-access<\/code><\/td>Select sandbox policy.<\/td><\/tr>
--search<\/code><\/td>Enable live web search instead of cached web search.<\/td><\/tr>
PROMPT<\/code><\/td>Start Codex with an initial instruction.<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n
Example:<\/p>\n\n\n\n
codex --cd apps\/frontend --model gpt-5.4 --sandbox workspace-write \"Review this app structure\"\n<\/code><\/pre>\n\n\n\n
7. Master list: codex<\/code> subcommands<\/h2>\n\n\n\n
Command<\/th>Status<\/th>Purpose<\/th><\/tr><\/thead>
codex<\/code><\/td>Stable<\/td>Launch the interactive terminal UI.<\/td><\/tr>
codex app<\/code><\/td>Stable<\/td>Launch Codex desktop app on macOS or Windows.<\/td><\/tr>
codex app-server<\/code><\/td>Experimental<\/td>Launch local Codex app server for development\/debugging.<\/td><\/tr>
codex apply<\/code> \/ codex a<\/code><\/td>Stable<\/td>Apply the latest diff from a Codex Cloud task.<\/td><\/tr>
codex cloud<\/code> \/ codex cloud-tasks<\/code><\/td>Experimental<\/td>Browse or execute Codex Cloud tasks from terminal.<\/td><\/tr>
codex completion<\/code><\/td>Stable<\/td>Generate shell completions.<\/td><\/tr>
codex debug app-server send-message-v2<\/code><\/td>Experimental<\/td>Debug app-server by sending one V2 message.<\/td><\/tr>
codex exec<\/code> \/ codex e<\/code><\/td>Stable<\/td>Run Codex non-interactively for scripts\/CI.<\/td><\/tr>
codex execpolicy<\/code><\/td>Experimental<\/td>Evaluate execpolicy rules.<\/td><\/tr>
codex features<\/code><\/td>Stable<\/td>List, enable, or disable feature flags.<\/td><\/tr>
codex fork<\/code><\/td>Stable<\/td>Fork a previous interactive session.<\/td><\/tr>
codex login<\/code><\/td>Stable<\/td>Authenticate with ChatGPT OAuth, device auth, or API key.<\/td><\/tr>
codex logout<\/code><\/td>Stable<\/td>Remove saved credentials.<\/td><\/tr>
codex mcp<\/code><\/td>Experimental<\/td>Manage MCP servers.<\/td><\/tr>
codex mcp-server<\/code><\/td>Experimental<\/td>Run Codex itself as an MCP server over stdio.<\/td><\/tr>
codex plugin marketplace<\/code><\/td>Experimental<\/td>Add, upgrade, or remove plugin marketplaces.<\/td><\/tr>
codex resume<\/code><\/td>Stable<\/td>Resume an earlier interactive session.<\/td><\/tr>
codex sandbox<\/code><\/td>Experimental<\/td>Run commands inside Codex-provided sandboxes.<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n
OpenAI\u2019s command reference says the maturity labels help distinguish Stable from Experimental\/Beta functionality, and the command reference catalogs the documented commands and flags. (OpenAI Developers<\/a>)<\/p>\n\n\n\n
8. Non-interactive automation with codex exec<\/code><\/h2>\n\n\n\n
For CI-style or scripted usage, use:<\/p>\n\n\n\n
codex exec \"Find and fix failing tests\"\n<\/code><\/pre>\n\n\n\n
Useful options include:<\/p>\n\n\n\n
Option<\/th>Purpose<\/th><\/tr><\/thead>
--json<\/code> \/ --experimental-json<\/code><\/td>Output newline-delimited JSON events.<\/td><\/tr>
--output-last-message, -o path<\/code><\/td>Write final assistant message to a file.<\/td><\/tr>
--output-schema path<\/code><\/td>Validate final response against a JSON schema.<\/td><\/tr>
--ephemeral<\/code><\/td>Avoid persisting session rollout files.<\/td><\/tr>
--skip-git-repo-check<\/code><\/td>Allow running outside a Git repo.<\/td><\/tr>
codex exec resume [SESSION_ID]<\/code><\/td>Resume a previous non-interactive session.<\/td><\/tr>
--last<\/code><\/td>Resume most recent session.<\/td><\/tr>
--all<\/code><\/td>Search sessions outside current working directory.<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n
OpenAI documents codex exec<\/code> as the automation-friendly command and notes that it can stream formatted text or JSONL and resume previous sessions. (OpenAI Developers<\/a>)<\/p>\n\n\n\n
9. Cloud task commands<\/h2>\n\n\n\n
Codex Cloud can be controlled from the terminal:<\/p>\n\n\n\n
codex cloud\n<\/code><\/pre>\n\n\n\n
Start a task directly:<\/p>\n\n\n\n
codex cloud exec --env ENV_ID --attempts 3 \"Summarize open bugs\"\n<\/code><\/pre>\n\n\n\n
List recent tasks:<\/p>\n\n\n\n
codex cloud list --env ENV_ID --limit 10 --json\n<\/code><\/pre>\n\n\n\n
Key options:<\/p>\n\n\n\n
Option<\/th>Purpose<\/th><\/tr><\/thead>
--env ENV_ID<\/code><\/td>Target cloud environment.<\/td><\/tr>
--attempts 1-4<\/code><\/td>Run best-of-N attempts.<\/td><\/tr>
--limit 1-20<\/code><\/td>Limit task list output.<\/td><\/tr>
--cursor string<\/code><\/td>Paginate cloud task list.<\/td><\/tr>
--json<\/code><\/td>Emit machine-readable JSON.<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n
The CLI features page says codex cloud<\/code> opens an interactive task picker, while codex cloud exec<\/code> submits a cloud task directly. (OpenAI Developers<\/a>)<\/p>\n\n\n\n
10. MCP commands<\/h2>\n\n\n\n
Codex can manage Model Context Protocol servers through:<\/p>\n\n\n\n
codex mcp\n<\/code><\/pre>\n\n\n\n
Common commands:<\/p>\n\n\n\n
Command<\/th>Purpose<\/th><\/tr><\/thead>
codex mcp list<\/code><\/td>List configured MCP servers.<\/td><\/tr>
codex mcp list --json<\/code><\/td>List servers as JSON.<\/td><\/tr>
codex mcp get <name><\/code><\/td>Show one server config.<\/td><\/tr>
codex mcp add <name> -- <command...><\/code><\/td>Add a stdio MCP server.<\/td><\/tr>
codex mcp add <name> --url <value><\/code><\/td>Add a streamable HTTP MCP server.<\/td><\/tr>
codex mcp login <name><\/code><\/td>Start OAuth login for supported HTTP servers.<\/td><\/tr>
codex mcp logout <name><\/code><\/td>Remove OAuth credentials.<\/td><\/tr>
codex mcp remove <name><\/code><\/td>Delete stored MCP server definition.<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n
codex mcp-server<\/code> is separate: it runs Codex itself as an MCP server over stdio so another tool can connect to Codex. (OpenAI Developers<\/a>)<\/p>\n\n\n\n
11. Resume, fork, and session management<\/h2>\n\n\n\n
Use:<\/p>\n\n\n\n
codex resume\ncodex resume --last\ncodex resume --all\ncodex resume <SESSION_ID>\n<\/code><\/pre>\n\n\n\n
Use:<\/p>\n\n\n\n
codex fork\ncodex fork --last\ncodex fork <SESSION_ID>\n<\/code><\/pre>\n\n\n\n
resume<\/code> continues a saved conversation; fork<\/code> branches a previous session into a new thread while preserving the original transcript. (OpenAI Developers<\/a>)<\/p>\n\n\n\n
12. Suggested beginner workflow<\/h2>\n\n\n\n
For daily coding:<\/p>\n\n\n\n
codex --cd path\/to\/repo\n<\/code><\/pre>\n\n\n\n
Inside Codex:<\/p>\n\n\n\n
\/status\n\/model\n\/plan\n\/diff\n\/review\n\/compact\n<\/code><\/pre>\n\n\n\n
For automation:<\/p>\n\n\n\n
codex exec --cd path\/to\/repo --json \"Run tests and summarize failures\"\n<\/code><\/pre>\n\n\n\n
For safer work, prefer read-only<\/code> or workspace-write<\/code> sandboxing. Use danger-full-access<\/code>, --yolo<\/code>, or approval bypassing only in isolated environments because OpenAI explicitly warns those modes bypass approval and sandbox protections. (OpenAI Developers<\/a>)<\/p>\n\n\n\n
13. Final cheat sheet<\/h2>\n\n\n\n
Use \/model<\/code> before complex tasks.
Use \/plan<\/code> before multi-file edits.
Use \/diff<\/code> and \/review<\/code> before commits.
Use \/compact<\/code> during long sessions.
Use \/clear<\/code> to start a new chat without quitting.
Use \/status<\/code> whenever you are unsure what model, permissions, or context state you are in.
Use codex exec<\/code> for scripting and CI.
Use codex cloud<\/code> for remote cloud tasks.
Use codex mcp<\/code> when Codex needs external tools.
Use skills instead of deprecated custom prompts for reusable workflows where possible.<\/p>\n","protected":false},"excerpt":{"rendered":"
OpenAI Codex has grown from a simple terminal coding assistant into a full workflow tool with terminal UI commands, desktop-app […]<\/p>\n","protected":false},"author":4,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[],"class_list":["post-2965","post","type-post","status-publish","format-standard","hentry","category-uncategorized"],"_links":{"self":[{"href":"https:\/\/aiopsschool.com\/blog\/wp-json\/wp\/v2\/posts\/2965","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/aiopsschool.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/aiopsschool.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/aiopsschool.com\/blog\/wp-json\/wp\/v2\/users\/4"}],"replies":[{"embeddable":true,"href":"https:\/\/aiopsschool.com\/blog\/wp-json\/wp\/v2\/comments?post=2965"}],"version-history":[{"count":1,"href":"https:\/\/aiopsschool.com\/blog\/wp-json\/wp\/v2\/posts\/2965\/revisions"}],"predecessor-version":[{"id":2966,"href":"https:\/\/aiopsschool.com\/blog\/wp-json\/wp\/v2\/posts\/2965\/revisions\/2966"}],"wp:attachment":[{"href":"https:\/\/aiopsschool.com\/blog\/wp-json\/wp\/v2\/media?parent=2965"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/aiopsschool.com\/blog\/wp-json\/wp\/v2\/categories?post=2965"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/aiopsschool.com\/blog\/wp-json\/wp\/v2\/tags?post=2965"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}