How do I install the Codex skill?

Run npx skillmds@latest add ashchupliak/codex in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Codex safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Codex?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Codex free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

ashchupliak (@ashchupliak) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Codex by ashchupliak — Agent Skill

Name: Codex
Author: ashchupliak

File contents

Codex Skill - JetBrains AI Integration

Use Codex CLI to delegate coding tasks that benefit from parallel AI processing.

When to Use Codex

GOOD for Codex:

Large-scale refactoring across many files
Repetitive code generation (boilerplate, tests)
Code analysis and documentation
Pattern-based transformations
Exploring unfamiliar codebases

KEEP in Claude:

Complex architectural decisions
Security-sensitive code
Tasks requiring deep context understanding
Interactive problem-solving with user

Configuration

Codex is pre-configured with JetBrains AI staging:

Provider: jbai-staging
Models: gpt-4o-2024-11-20, o1-2024-12-17, o3-mini-2025-01-31

Usage Patterns

Quick Analysis (read-only)

codex exec -c model_provider=jbai-staging --model "gpt-4o-2024-11-20" \
  --sandbox read-only \
  "Analyze the codebase and list all REST endpoints" 2>/dev/null

Code Generation (workspace-write)

codex exec -c model_provider=jbai-staging --model "gpt-4o-2024-11-20" \
  --sandbox workspace-write --full-auto \
  "Generate unit tests for UserService.kt" 2>/dev/null

Heavy Reasoning (o1/o3 models)

codex exec -c model_provider=jbai-staging --model "o3-mini-2025-01-31" \
  --sandbox read-only \
  "Review security vulnerabilities in auth module" 2>/dev/null

Resume Previous Session

echo "continue with the refactoring" | codex exec --skip-git-repo-check resume --last 2>/dev/null

Decision Framework

Before delegating to Codex, consider:

Factor	Use Codex	Keep in Claude
File count	Many files	Few files
Task type	Repetitive/mechanical	Creative/strategic
Context needed	Self-contained	Requires conversation context
User interaction	None needed	Back-and-forth required

Execution Protocol

Announce intent: Tell user you're delegating to Codex and why
Choose model:
- gpt-4o-2024-11-20 for fast tasks
- o1-2024-12-17 / o3-mini-2025-01-31 for complex reasoning
Choose sandbox:
- read-only for analysis
- workspace-write for edits
Run command with 2>/dev/null to suppress thinking tokens
Report results back to user
Offer resume: "Say 'codex resume' to continue this session"

Safety Rules

Always use --skip-git-repo-check flag
Default to read-only sandbox unless edits needed
Never use danger-full-access without explicit user permission
Suppress stderr with 2>/dev/null by default

---name: codexdescription: Delegate coding tasks to Codex CLI (OpenAI gpt-5-codex via JetBrains AI). Use when a task involves repetitive code generation, refactoring, or analysis that can be offloaded.---# Codex Skill - JetBrains AI IntegrationUse Codex CLI to delegate coding tasks that benefit from parallel AI processing.## When to Use Codex**GOOD for Codex:**- Large-scale refactoring across many files- Repetitive code generation (boilerplate, tests)- Code analysis and documentation- Pattern-based transformations- Exploring unfamiliar codebases**KEEP in Claude:**- Complex architectural decisions- Security-sensitive code- Tasks requiring deep context understanding- Interactive problem-solving with user## ConfigurationCodex is pre-configured with JetBrains AI staging:- Provider: `jbai-staging`- Models: `gpt-4o-2024-11-20`, `o1-2024-12-17`, `o3-mini-2025-01-31`## Usage Patterns### Quick Analysis (read-only)```bashcodex exec -c model_provider=jbai-staging --model "gpt-4o-2024-11-20" \ --sandbox read-only \ "Analyze the codebase and list all REST endpoints" 2>/dev/null```### Code Generation (workspace-write)```bashcodex exec -c model_provider=jbai-staging --model "gpt-4o-2024-11-20" \ --sandbox workspace-write --full-auto \ "Generate unit tests for UserService.kt" 2>/dev/null```### Heavy Reasoning (o1/o3 models)```bashcodex exec -c model_provider=jbai-staging --model "o3-mini-2025-01-31" \ --sandbox read-only \ "Review security vulnerabilities in auth module" 2>/dev/null```### Resume Previous Session```bashecho "continue with the refactoring" | codex exec --skip-git-repo-check resume --last 2>/dev/null```## Decision FrameworkBefore delegating to Codex, consider:| Factor | Use Codex | Keep in Claude ||--------|-----------|----------------|| File count | Many files | Few files || Task type | Repetitive/mechanical | Creative/strategic || Context needed | Self-contained | Requires conversation context || User interaction | None needed | Back-and-forth required |## Execution Protocol1. **Announce intent**: Tell user you're delegating to Codex and why2. **Choose model**: - `gpt-4o-2024-11-20` for fast tasks - `o1-2024-12-17` / `o3-mini-2025-01-31` for complex reasoning3. **Choose sandbox**: - `read-only` for analysis - `workspace-write` for edits4. **Run command** with `2>/dev/null` to suppress thinking tokens5. **Report results** back to user6. **Offer resume**: "Say 'codex resume' to continue this session"## Safety Rules- Always use `--skip-git-repo-check` flag- Default to `read-only` sandbox unless edits needed- Never use `danger-full-access` without explicit user permission- Suppress stderr with `2>/dev/null` by default

Frequently asked questions

Codex