From research-skills
Enables Claude Code and Codex to consult each other via one-shot commands or tmux sessions for second opinions, code review, or task dispatch.
How this skill is triggered — by the user, by Claude, or both
Slash command
/research-skills:conferThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Call the other agent. If you're Claude Code, call Codex. If you're Codex, call Claude Code.
Call the other agent. If you're Claude Code, call Codex. If you're Codex, call Claude Code.
For quick questions, reviews, or dispatched tasks.
Claude Code → Codex:
codex exec --dangerously-bypass-approvals-and-sandbox "<prompt>"
Codex → Claude Code:
claude -p --dangerously-skip-permissions "<prompt>"
Both return output to stdout. Use this for:
For extended collaboration where back-and-forth is needed.
tmux new-session -d -s confer -c "$(pwd)"
Send a message (two steps — Enter must be separate):
tmux send-keys -t confer "<command or prompt>" ""
tmux send-keys -t confer Enter
Read the response:
tmux capture-pane -t confer -p -S -50
Start the other agent in the session, then send follow-up prompts with send-keys. This mode suits open-ended collaboration: iterating on a design, working through a problem together, or anything that needs multiple exchanges.
One-shot — You have a clear question or task. Fire and forget.
Tmux — You need a conversation. Multiple rounds, evolving context, genuine collaboration.
Provides behavioral guidelines to reduce common LLM coding mistakes, focusing on simplicity, surgical changes, assumption surfacing, and verifiable success criteria.
Searches, retrieves, and installs Agent Skills from prompts.chat registry using MCP tools like search_skills and get_skill. Activates for finding skills, browsing catalogs, or extending Claude.
npx claudepluginhub cailmdaley/skills --plugin research-skills