research-lead-sidecar — Claude Code skill for agent delegation
The research-lead-sidecar skill keeps the lead agent on the critical path while bounded side tasks go to subagents. Use it for research-led work where the lead still needs to own the final decisions.
no license ★ 475 updated 2026-10-13
npx skills add cnfjlhj/ai-collab-playbook --skill research-lead-sidecar -a claude-codeWhat research-lead-sidecar does
The lead agent frames the work, combines findings, and makes the final calls. Subagents can take on separate research, checking, or writing tasks when their work is easy to inspect and does not block the main thread. The lead keeps moving while they work.
Agree on the delegation plan before launching subagents, with each role and task scope spelled out. Writing tasks need clearly separated file ownership; when scopes overlap, subagents should stay read-only. For tiny tasks or debugging that depends on each prior step, delegation adds handoff overhead. The lead decides whether to accept, adapt, or reject each result.
What it helps with
- split research work between a lead agent and subagents
- delegate a bounded check without pausing the main task
- define task scopes and file ownership for subagents
- review returned work and decide what to integrate
Example prompts for the agent
- Divide this research task between me as lead and one or two subagents; propose their roles and scopes, then wait for my approval.
- Resume the current lead-and-sidecar workflow, and do not start a duplicate subagent if one is still active.
- Give a subagent a separate verification task while you continue the critical path, then assess its findings.
How to install research-lead-sidecar
1Skills CLI (Vercel)
npx skills add cnfjlhj/ai-collab-playbook --skill research-lead-sidecar -a claude-code2Manually: put the files in place
~/.claude/skills/research-lead-sidecar/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/research-lead-sidecar/
1Skills CLI (Vercel)
npx skills add cnfjlhj/ai-collab-playbook --skill research-lead-sidecar -a codex2Manually: put the files in place
.agents/skills/research-lead-sidecar/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add cnfjlhj/ai-collab-playbook --skill research-lead-sidecar -a cursor2Manually: put the files in place
.cursor/skills/research-lead-sidecar/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install cnfjlhj/ai-collab-playbook skills/full/research-lead-sidecar2Manually: put the files in place
.github/skills/research-lead-sidecar/SKILL.md1Skills CLI (Vercel)
npx skills add cnfjlhj/ai-collab-playbook --skill research-lead-sidecar -a gemini-cli2Platform CLI
gemini skills install https://github.com/cnfjlhj/ai-collab-playbook --path skills/full/research-lead-sidecarManually: put the files in place
.devin/skills/research-lead-sidecar/SKILL.mdThe legacy .windsurf/skills/ path still works
Claude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about research-lead-sidecar
How do I install research-lead-sidecar in Claude Code?
Run npx skills add cnfjlhj/ai-collab-playbook --skill research-lead-sidecar -a claude-code. The skill file is at ~/.claude/skills/research-lead-sidecar/SKILL.md.
What does research-lead-sidecar do?
It guides a lead agent in delegating bounded side tasks while keeping responsibility for the main work, synthesis, and final decisions.
Should the lead get approval before spawning subagents?
Yes. The lead first proposes roles and scopes in an execution plan and waits for approval. After that, ordinary delegations do not need separate approval each time.