collaborating-with-claude-code
Delegate code implementation/review/debugging/alternatives to Claude Code via a JSON bridge script. Multi-turn via SESSION_ID.
install
source · Clone the upstream repo
git clone https://github.com/ZhenHuangLab/collaborating-with-claude-code
Claude Code · Install into ~/.claude/skills/
git clone --depth=1 https://github.com/ZhenHuangLab/collaborating-with-claude-code ~/.claude/skills/zhenhuanglab-collaborating-with-claude-code-collaborating-with-claude-code
manifest:
SKILL.mdsource content
Collaborating with Claude Code
Use this skill for second opinions, code review, test design, or code implementations/alternatives. The bridge runs
claude (Claude Code) non-interactively ("print" mode) and returns JSON.
The script is located at
~/.codex/skills/collaborating-with-claude-code/scripts/claude_code_bridge.py.
Timing
Claude Code often needs 1–2+ minutes per task.
- Prefer running the bridge directly (no
); increase&
as needed (default: 1800s).--timeout-s - Do NOT redirect stdout to a file (e.g.
).> /tmp/out.json - By default, the bridge streams Claude's assistant text to
, and prints only the final JSON envelope tostderr
.stdout
Context
- Do NOT read the script unless you are modifying it;
- Before running the script, ALWAYS use
to get the usage instructions.python <script_loc> --help
Usage
- please always require claude code to fully understand the codebase before responding or making any changes.
- Put collaborating-with-claude-code terminal commands in the background terminal.
- Always review claude code's responses (or changes it makes) and make sure they are correct, constructive and complete.
- When claude code asks clarifying questions in a multi-turn session, always respond to its questions in that session based on current situation.
Default
- full access (
): use only in trusted repos/directories.--full-access - extended thinking ON (can disable via
).--no-extended-thinking - step mode AUTO (can disable via
).--step-mode off
Output format
The bridge prints JSON to
stdout:
{"success": true, "SESSION_ID": "abc123", "agent_messages": "…Claude output…"}
Recommended delegation patterns
- Guided coding: "Implement the code for [feature] following these specific steps/constraints."
- Second opinion: "Propose an alternative approach and tradeoffs."
- Code review: "Find bugs, race conditions, security issues; propose fixes."
- Test design: "Write a test plan + edge cases; include example test code."
- Diff review: "Review this patch; point out regressions and missing cases."