Awesome Agent Config workflow
Set up Claude Code and Codex in one conversation: global instructions, rules, plugins, skills, a status line and correction memory
- 260 ★ on the repository
- Last commit
- MIT
Broad access
Why this level
- The agent edits global Claude Code and Codex instructions and settings
- A separate template removes Codex approval prompts and sandbox restrictions; Claude’s option grants broad command and file permissions
- Installation pulls third-party skills and plugins from their repositories
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add workflows/awesome-agent-configDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- The agent edits global Claude Code and Codex instructions and settings
- A separate template removes Codex approval prompts and sandbox restrictions; Claude’s option grants broad command and file permissions
- Installation pulls third-party skills and plugins from their repositories
This is third-party code. Review the repository files before installing.
What it does
Awesome Agent Config is a repository for configuring Claude Code and Codex through an agent. Open the checkout or share its URL and ask the agent to read `INSTALL.md`. It determines the OS, client and current configuration, then presents categorized options with descriptions and installation status. The instructions call for a choice, a preview and installation of agreed items. The catalog covers global instructions, base settings, permissions, writing and programming rules, review tools, workflows, a status line and a lessons journal. Claude’s selection includes `adversarial-review` with skeptic, architect and minimalist roles. Codex has a different selection; its retained adversarial-review files are historical and are not installed. Third-party components come from their own sources. Later requests such as adding or removing an item are handled through `edit-config`. Skipping an item during an update does not remove it, and previous choices are retained. Windows and WSL are detected and configured separately.
Who it is for. For people who use Claude Code and Codex and want to assemble a shared working configuration in one conversation and then edit it with short requests.
Good fit when
- You just installed Claude Code or Codex and want working instructions, rules and a status line without manual assembly
- You need the same base settings for two agents but independent instructions and memory for each
- You work on Windows with WSL and need both environments configured separately
- You want to add review skills such as adversarial-review and see what from the catalogue is already installed
- Later you need to add or remove a single item with an "add" or "remove" request
Not a fit when
- You use another agent: the catalogue targets Claude Code and Codex
- You require manual configuration without an agent modifying global files
- The current setup must stay untouched: even a careful install puts the chosen templates into the agent configuration
- You need a single skill from the catalogue: it is easier to install from its source directly
Example request
Read INSTALL.md from this repository and configure Claude Code: show numbered options by category and install what I pickLimitations
Installation follows the text of `INSTALL.md`, so preservation of settings depends on the agent’s actions. Preview, backups and partial merging are prescribed; inspect proposed changes before writing. Claude Code and Codex selections differ.
Any selected Codex item requires disabling `desktop.external-agent-import-sync-enabled`, even without base settings selected. This does not remove existing imports. A separate high-permission option sets Codex to `approval_policy=never` and `sandbox_mode=danger-full-access`. In Claude it enables automatic mode and broad shell, read and write permissions.
These settings weaken client restrictions. Third-party plugins, MCP servers and APIs retain their own permissions and terms. Repository-owned materials use MIT; the listed lieflat-charts component allows noncommercial use only under PolyForm Noncommercial 1.0.0.
How to disable. Ask `edit-config` to remove a selected item, first checking ownership and local edits. Remove native plugins through the client and edit mixed settings selectively. Skipping an item during an update does not remove it. MCP authorizations, fonts and installed packages also need separate inspection.
FAQ
How do I set up Claude Code and Codex with Awesome Agent Config?
Open the repository in your agent or share its page URL and ask it to read INSTALL.md and configure the current agent. The agent lists options in the chat, you pick numbers, and it installs the selection.
What does the edit-config skill do?
It serves configuration requests: show, add, edit, remove, repair and update items. Read-only queries change nothing, and the skill tracks this repository's main branch.
What is adversarial-review in this repository?
A cross-model review skill that examines changes through the roles of skeptic, architect and minimalist. The original idea comes from poteto/noodle, and Awesome Agent Config ships an adapted version for Claude.
Does it work on Windows?
Yes. The repository states support for macOS, Linux, Windows and WSL. Windows and WSL are detected and configured separately, and if App and CLI share a home directory the installation is reused.
Will the agent install everything at once?
The instructions provide item selection and a preview. Codex has one mandatory prerequisite outside that selection: automatic external-agent import is disabled before installing any item. High permissions remain a separate, explicit choice.
Discussion
No questions or reviews yet.
Related
WorkflowsNotes, tasks, contacts and project context live in Markdown on your disk, and Claude Code and Codex read the same memory
cafe3310's personal skill collection: spec-driven project work, reviews, reports, browser, 3D and video in one repository
Agent Standup shows Claude Code, Codex and Mistral Vibe sessions as a live team call: who works, who waits.
A library of 229 skills for work and life: sales, marketing, finance, engineering, design and parallel agent runs
A local web editor for Claude Code config: shows project sizes, deletes old chat histories and manages MCP servers
A workflow for Claude Code, Codex and Gemini CLI where the agent files source documents into a persistent, linked wiki
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| Awesome Agent Config (this card) | no | 260 | |
| ZCF | no | 6,071 | installs workflows and MCP with one console command, while this repository configures through a conversation with the agent |
| Agent Scripts | no | 7,463 | one author's personal set of shared rules and skills |
| Agent Purge | no | 1,299 | A CLI switches Markdown instruction profiles across coding agents and shows active integrations in a local dashboard |
README in short
Component selections and revisions are tracked in `selection.json`; file ownership, hashes and backups are recorded in `files.json`. These records help distinguish managed materials from user content during subsequent changes.