One setup
Every AI CLI

Keep agents, skills, rules, hooks, and MCP servers in one portable source. agnostic-ai writes each tool's native files and checks them for drift.

Install and start Try in the browser

Recommended installerAll install options

curl -fsSL https://raw.githubusercontent.com/Chemaclass/agnostic-ai/main/scripts/install.sh | bash

25 targets. 10 portable spec kinds. Byte-stable output.

Set up agnostic-ai with a coding agent

Your setup outlives the tool you picked.

Every AI coding tool ships its own config file. Adopt three and your instructions live in three places, drifting apart. agnostic-ai keeps one source in your repository and writes the rest.

You own the source

Plain Markdown and YAML in your repo. No account, no service. import brings an existing setup in.

Generated files are outputs

Same input, same bytes, so they diff cleanly. sync --check fails CI when they drift.

No tool is privileged

Adding a target never changes what the others get. One rule routes to every native path and scope syntax.

A sync layer, not a platform

Easy to stop using. The generated files are yours, and they keep working.

Why not just symlink one file? Each tool reads a different format at a different path. A symlink shares bytes. It cannot translate them. See how the formats diverge

Write it once. Every tool reads it natively.

Everything lives in .agnostic-ai/, next to agnostic-ai.yaml, which picks your targets. Pick an entry to see a source file and the native files sync writes from it, each in its own tool's format.

.agnostic-ai/

Also commands/, reviews/, environments/, and ignore/. See every spec kind.

Project instructions. Each tool gets the same body in its own entry file.

.agnostic-ai/AGNOSTIC_AI.md

# Billing API

- Run `pnpm test` before every commit.
- Never edit `prisma/migrations/` by hand.

sync

CLAUDE.md

# Billing API

- Run `pnpm test` before every commit.
- Never edit `prisma/migrations/` by hand.

AGENTS.md

# Billing API

- Run `pnpm test` before every commit.
- Never edit `prisma/migrations/` by hand.

<!-- agnostic-ai:rules:start -->

## Rules

### testing

<!-- source: .agnostic-ai/rules/testing.md -->
One behavior per test. Mock HTTP with `msw`.


### my-style

<!-- source: .agnostic-ai/local/rules/my-style.md -->
Answer in short sentences.


<!-- agnostic-ai:rules:end -->

GEMINI.md

# Billing API

- Run `pnpm test` before every commit.
- Never edit `prisma/migrations/` by hand.

<!-- agnostic-ai:rules:start -->

## Rules

### testing

<!-- source: .agnostic-ai/rules/testing.md -->
One behavior per test. Mock HTTP with `msw`.


### my-style

<!-- source: .agnostic-ai/local/rules/my-style.md -->
Answer in short sentences.


<!-- agnostic-ai:rules:end -->

Docs: Entry-point files

Subagents with a role, their tools, and a model per tool.

.agnostic-ai/agents/test-writer.md

---
name: test-writer
description: Writes a failing test for a bug.
tools: [Read, Grep, Edit]
model:
  claude: claude-sonnet-5
  codex: gpt-6-sol
effort: medium
---

Reproduce the bug in a failing test. Do not touch production code.

sync

.claude/agents/test-writer.md

---
name: test-writer
description: Writes a failing test for a bug.
tools:
  - Read
  - Grep
  - Edit
model: claude-sonnet-5
effort: medium
---

Reproduce the bug in a failing test. Do not touch production code.

.codex/agents/test-writer.toml

name = "test-writer"
description = "Writes a failing test for a bug."
model = "gpt-6-sol"
model_reasoning_effort = "medium"
developer_instructions = """
Reproduce the bug in a failing test. Do not touch production code.
"""

.gemini/agents/test-writer.md

---
name: test-writer
description: Writes a failing test for a bug.
tools:
  - read_file
  - grep_search
  - replace
---

Reproduce the bug in a failing test. Do not touch production code.

Saved prompts you run by name.

.agnostic-ai/commands/review.md

---
name: review
description: Review a change against main.
targets: [kiro]
---

Review $ARGUMENTS against main and report bugs with file and line references.

sync

.kiro/prompts/review.md

Review $ARGUMENTS against main and report bugs with file and line references.

Kiro CLI V3 reads workspace prompts as slash commands.

Commands that run on tool events, such as after every edit.

.agnostic-ai/hooks/format.yaml

name: format
targets: [claude, codex]
event: PostToolUse
matcher: Edit|Write
command: .claude/hooks/format.sh

sync

.claude/settings.json

{
  "statusLine": {
    "type": "command",
    "command": "~/.claude/statusline.sh"
  },
  "permissions": {
    "allow": [
      "Bash(pnpm test:*)"
    ],
    "ask": [
      "Bash(git push:*)"
    ],
    "deny": [
      "Read(.env)"
    ]
  },
  "model": "claude-opus-5-5",
  "effortLevel": "xhigh",
  "hooks": {
    "PostToolUse": [
      {
        "matcher": "Edit|Write",
        "hooks": [
          {
            "type": "command",
            "command": ".claude/hooks/format.sh"
          }
        ]
      }
    ]
  },
  "env": {
    "AGNOSTIC_AI_TARGET": "claude"
  }
}

.codex/hooks.json

{
  "hooks": {
    "PostToolUse": [
      {
        "matcher": "Edit|Write",
        "hooks": [
          {
            "type": "command",
            "command": "export AGNOSTIC_AI_TARGET=codex; .codex/hooks/format.sh",
            "commandWindows": ".codex/hooks/format.sh"
          }
        ]
      }
    ]
  }
}

Events are written as is. Claude Code and Codex share PostToolUse; Gemini CLI names its events its own way, such as AfterTool, so it takes a hook spec of its own.

Your own specs on this machine, ignored by Git. Sync layers them over the shared specs.

.agnostic-ai/local/rules/my-style.md

---
name: my-style
alwaysApply: true
---

Answer in short sentences.

sync

.claude/rules/my-style.md

---
name: my-style
---

Answer in short sentences.

AGENTS.md

# Billing API

- Run `pnpm test` before every commit.
- Never edit `prisma/migrations/` by hand.

<!-- agnostic-ai:rules:start -->

## Rules

### testing

<!-- source: .agnostic-ai/rules/testing.md -->
One behavior per test. Mock HTTP with `msw`.


### my-style

<!-- source: .agnostic-ai/local/rules/my-style.md -->
Answer in short sentences.


<!-- agnostic-ai:rules:end -->

GEMINI.md

# Billing API

- Run `pnpm test` before every commit.
- Never edit `prisma/migrations/` by hand.

<!-- agnostic-ai:rules:start -->

## Rules

### testing

<!-- source: .agnostic-ai/rules/testing.md -->
One behavior per test. Mock HTTP with `msw`.


### my-style

<!-- source: .agnostic-ai/local/rules/my-style.md -->
Answer in short sentences.


<!-- agnostic-ai:rules:end -->

Docs: Local overrides

MCP server connections, one YAML file per server. Secrets stay in environment variables.

.agnostic-ai/mcps/github.yaml

name: github
type: http
url: https://api.githubcopilot.com/mcp/
headers:
  Authorization: Bearer ${GITHUB_TOKEN}

sync

.mcp.json

{
  "mcpServers": {
    "github": {
      "headers": {
        "Authorization": "Bearer ${GITHUB_TOKEN}"
      },
      "type": "http",
      "url": "https://api.githubcopilot.com/mcp/"
    }
  }
}

.codex/config.toml

model = "gpt-6-sol"
model_reasoning_effort = "high"

[tui]
notification_condition = "always"

[mcp_servers.github]
url = "https://api.githubcopilot.com/mcp/"
bearer_token_env_var = "GITHUB_TOKEN"

.gemini/settings.json

{
  "mcpServers": {
    "github": {
      "headers": {
        "Authorization": "Bearer ${GITHUB_TOKEN}"
      },
      "httpUrl": "https://api.githubcopilot.com/mcp/"
    }
  }
}

Docs: MCP servers

Native keys no spec covers, one file per tool: the Claude Code status line, Codex notification settings. Sync merges each into that tool's own file.

.agnostic-ai/overlays/claude.settings.json

{
  "statusLine": {
    "type": "command",
    "command": "~/.claude/statusline.sh"
  }
}

.agnostic-ai/overlays/codex.config.toml

[tui]
notification_condition = "always"

sync

.claude/settings.json

{
  "statusLine": {
    "type": "command",
    "command": "~/.claude/statusline.sh"
  },
  "permissions": {
    "allow": [
      "Bash(pnpm test:*)"
    ],
    "ask": [
      "Bash(git push:*)"
    ],
    "deny": [
      "Read(.env)"
    ]
  },
  "model": "claude-opus-5-5",
  "effortLevel": "xhigh",
  "hooks": {
    "PostToolUse": [
      {
        "matcher": "Edit|Write",
        "hooks": [
          {
            "type": "command",
            "command": ".claude/hooks/format.sh"
          }
        ]
      }
    ]
  },
  "env": {
    "AGNOSTIC_AI_TARGET": "claude"
  }
}

.codex/config.toml

model = "gpt-6-sol"
model_reasoning_effort = "high"

[tui]
notification_condition = "always"

[mcp_servers.github]
url = "https://api.githubcopilot.com/mcp/"
bearer_token_env_var = "GITHUB_TOKEN"

Docs: Claude settings

Conventions, always on or tied to file patterns and folders.

.agnostic-ai/rules/testing.md

---
name: testing
globs: ["**/*.test.ts"]
---

One behavior per test. Mock HTTP with `msw`.

sync

.claude/rules/testing.md

---
name: testing
paths:
  - "**/*.test.ts"
---

One behavior per test. Mock HTTP with `msw`.

AGENTS.md

# Billing API

- Run `pnpm test` before every commit.
- Never edit `prisma/migrations/` by hand.

<!-- agnostic-ai:rules:start -->

## Rules

### testing

<!-- source: .agnostic-ai/rules/testing.md -->
One behavior per test. Mock HTTP with `msw`.


### my-style

<!-- source: .agnostic-ai/local/rules/my-style.md -->
Answer in short sentences.


<!-- agnostic-ai:rules:end -->

GEMINI.md

# Billing API

- Run `pnpm test` before every commit.
- Never edit `prisma/migrations/` by hand.

<!-- agnostic-ai:rules:start -->

## Rules

### testing

<!-- source: .agnostic-ai/rules/testing.md -->
One behavior per test. Mock HTTP with `msw`.


### my-style

<!-- source: .agnostic-ai/local/rules/my-style.md -->
Answer in short sentences.


<!-- agnostic-ai:rules:end -->

Hook script bodies. Sync copies each one into the hooks folder of every tool the hook targets.

.agnostic-ai/scripts/claude/format.sh

#!/bin/sh
file=$(jq -r '.tool_input.file_path // empty')
[ -z "$file" ] || npx prettier --write "$file"

sync

.claude/hooks/format.sh

#!/bin/sh
file=$(jq -r '.tool_input.file_path // empty')
[ -z "$file" ] || npx prettier --write "$file"

.codex/hooks/format.sh

#!/bin/sh
file=$(jq -r '.tool_input.file_path // empty')
[ -z "$file" ] || npx prettier --write "$file"

The default model and effort per tool, and permissions.

.agnostic-ai/settings/model.yaml

model:
  claude: claude-opus-5-5
  codex: gpt-6-sol
effort:
  claude: xhigh
  codex: high

.agnostic-ai/settings/permissions.yaml

permissions:
  allow: [Bash(pnpm test:*)]
  ask: [Bash(git push:*)]
  deny: [Read(.env)]

sync

.claude/settings.json

{
  "statusLine": {
    "type": "command",
    "command": "~/.claude/statusline.sh"
  },
  "permissions": {
    "allow": [
      "Bash(pnpm test:*)"
    ],
    "ask": [
      "Bash(git push:*)"
    ],
    "deny": [
      "Read(.env)"
    ]
  },
  "model": "claude-opus-5-5",
  "effortLevel": "xhigh",
  "hooks": {
    "PostToolUse": [
      {
        "matcher": "Edit|Write",
        "hooks": [
          {
            "type": "command",
            "command": ".claude/hooks/format.sh"
          }
        ]
      }
    ]
  },
  "env": {
    "AGNOSTIC_AI_TARGET": "claude"
  }
}

.codex/config.toml

model = "gpt-6-sol"
model_reasoning_effort = "high"

[tui]
notification_condition = "always"

[mcp_servers.github]
url = "https://api.githubcopilot.com/mcp/"
bearer_token_env_var = "GITHUB_TOKEN"

Codex and Gemini CLI have no portable permission list, and the model map names no Gemini model, so sync prints a note for each instead of guessing.

Reusable workflows. Files beside SKILL.md, such as check.sh, copy with it.

.agnostic-ai/skills/release/SKILL.md

---
name: release
description: Cut a release and verify it.
---

Run `./check.sh`, bump the version, and push the tag.

sync

.claude/skills/release/SKILL.md

---
name: release
description: Cut a release and verify it.
---

Run `./check.sh`, bump the version, and push the tag.

.agents/skills/release/SKILL.md

---
name: release
description: Cut a release and verify it.
---

Run `./check.sh`, bump the version, and push the tag.

.gemini/skills/release/SKILL.md

---
name: release
description: Cut a release and verify it.
---

Run `./check.sh`, bump the version, and push the tag.

Four commands. Install, import, preview, sync.

Install

npm install -g agnostic-ai

Node 18 or newer, from your project root.

Start or import

agnostic-ai init --from all

Import existing AI tool files first.

Review the output

agnostic-ai sync --plan

Review the planned changes before writing them.

Write the files

agnostic-ai sync

Write every native file for the targets you chose.

All install options Open the getting started guide Migrate an existing setup

Keep the setup current. Know what changed.

What shipped in agnostic-ai, and what changed upstream in the tools it targets.

Read the latest briefing Browse all updates Subscribe by RSS

Latest briefings

06 October 2026

agnostic-ai v0.80.0: session handoffs between tools

05 October 2026

agnostic-ai v0.79.0: portable hook guards and tool permissions

03 October 2026

agnostic-ai v0.78.0: hook run for six more tools

Bring your existing setup.
Keep every target in sync.

agnostic-ai init --from all

Read getting started Set up with an agent