What does this PR do?

Codex is supporting subagents now.
https://developers.openai.com/codex/subagents

Adds Codex support to agency-agents as a standard converted integration.

This change:

• adds codex to scripts/convert.sh
• adds codex to scripts/install.sh
• generates Codex custom agent TOML files under integrations/codex/agents/
• installs generated Codex agents to ~/.codex/agents/
• documents Codex usage in the main README, integrations README, and a new integrations/codex/README.md

The Codex mapping is intentionally minimal and keeps the source agents as close to 1:1 as possible:

• name comes from source frontmatter
• description comes from source frontmatter
• developer_instructions contains the source Markdown body

Unsupported source-only metadata such as color, emoji, and vibe is omitted.

The converter also escapes TOML string content safely so existing agents with control characters in the Markdown body do not generate invalid Codex agent files.

Verification

Verified locally by:

• running ./scripts/convert.sh --tool codex
• running ./scripts/install.sh --tool codex
• confirming installed agents were written to ~/.codex/agents/
• parsing all generated/installed Codex TOML files successfully
• confirming codex works with the installed custom agents in a live local CLI run

Notes

• Codex is added as a converted integration, similar to other non-native tool targets in this repo.
• Generated Codex artifacts are ignored via .gitignore.

Checklist

• Follows the agent template structure from CONTRIBUTING.md
• Includes YAML frontmatter with name, description, color
• Tested in real scenarios
• Proofread and formatted correctly