cc-doubleteam
github.com/responsiblparty/cc-doubleteam →Two AIs, one workflow — one thinks, one ships
WHAT IT SOLVES
One AI playing architect, coder, and reviewer at the same time is asking for trouble
WHY IT'S INTERESTING
Each model does what it's actually good at
Claude plans and reviews. Codex just executes. Not a random API sandwich — it's assigning roles by actual model strengths. Claude reasons, Codex codes
The commit history tells the real story
The author first tried invoking Codex as a subagent. Didn't work. Switched to direct Bash invocation. That 'fix:' commit message is the kind of thing you only write after actually hitting a wall
「Claude plans, Codex executes, Claude reviews」
TECH GUESS
Claude Code skills system orchestrating Codex CLI via Bash — lightweight scripting, no heavy framework
Discussion (0)
- No comments yet — be the first.
Related
#157▶ 183Rate-limited on Claude? Switch rooms and keep going
#155▶ 149Your AI agent made 12 HTML files. Now what?
#154▶ 141Your AI agent won't tell you it broke your MCP server. Spanly will.