Codex

LNAI exports unified configuration to Codex Team Config under the .codex/ directory.

Output Structure

AGENTS.md            # Symlink -> .ai/AGENTS.md (at project root)

.codex/
├── config.toml      # Generated (MCP servers)
├── skills/<name>/   # Symlinks -> ../../.ai/skills/<name>/
└── <overrides>      # Symlinks from .ai/.codex/

<dir>/AGENTS.md      # Generated (rules per directory)

File Mapping

Source	Output	Type
`.ai/AGENTS.md`	`AGENTS.md`	Symlink
`.ai/rules/*.md`	`<dir>/AGENTS.md`	Generated
`.ai/skills/<name>/`	`.codex/skills/<name>`	Symlink
`.ai/settings.json`	`.codex/config.toml`	Generated
`.ai/.codex/<path>`	`.codex/<path>`	Symlink

Generated config.toml

MCP servers are exported into Codex’s mcp_servers blocks:

// Input (LNAI format)
{
  "mcpServers": {
    "db": {
      "command": "npx",
      "args": ["-y", "@example/db"],
      "env": { "DB_URL": "${DB_URL}" }
    },
    "remote": {
      "url": "https://mcp.example.com",
      "headers": { "Authorization": "Bearer ${API_KEY}" }
    }
  }
}

# Output (Codex format)
[mcp_servers.db]
command = "npx"
args = ["-y", "@example/db"]
env = { DB_URL = "${DB_URL}" }

[mcp_servers.remote]
url = "https://mcp.example.com"
http_headers = { Authorization = "Bearer ${API_KEY}" }

Rules

Rules are grouped by directory using their paths globs and written to <dir>/AGENTS.md files. Only subdirectory rules are exported; rules targeting project root (.) are skipped and reported as warnings.

Unsupported Features

Permissions

Codex permissions rules are not generated from LNAI permissions. If your .ai/settings.json contains a permissions block, LNAI will display a “skipped” notice but continue syncing other configuration.

Overrides

Place Codex-specific files in .ai/.codex/ to have them symlinked directly:

.ai/.codex/
└── custom-config.toml -> .codex/custom-config.toml

When an override file exists at the same path as a generated file (e.g., config.toml), the override takes priority.