BioClaw bio-human-feedback

bio-human-feedback

install
source · Clone the upstream repo
git clone https://github.com/Runchuan-BU/BioClaw
Claude Code · Install into ~/.claude/skills/
T=$(mktemp -d) && git clone --depth=1 https://github.com/Runchuan-BU/BioClaw "$T" && mkdir -p ~/.claude/skills && cp -r "$T/container/skills/bio-human-feedback" ~/.claude/skills/runchuan-bu-bioclaw-bio-human-feedback && rm -rf "$T"
manifest: container/skills/bio-human-feedback/SKILL.md
source content

bio-human-feedback

Phase 2.6: Human review checkpoint (人类反馈验证)

Present the refined proposal to a human reviewer, collect feedback, and decide whether to continue or loop back for revision.

Purpose

  1. Present the current proposal in a concise reviewable form
  2. Wait for explicit human feedback
  3. Record approval or requested changes
  4. Route the workflow back to the correct phase if needed

Inputs

  • FINAL_PROPOSAL.md
  • current round number

Workflow

  1. Present the proposal summary
  2. Wait for approval or feedback
  3. Classify the feedback severity
  4. Record the response and decide the next phase

Human Review Summary Format

# Proposal Review - Round X

## Innovation Summary
- ...

## Figure Overview
- Figure 1:
- Figure 2:

## Experimental Plan
| Task | Dataset | Metric | Baseline |
|------|---------|--------|----------|
| ... | ... | ... | ... |

## Key Decisions Requiring Human Approval
1. ...
2. ...

## Reviewer Summary
| Reviewer | Score / status | Main advice |
|----------|----------------|-------------|
| Editor | ... | ... |
| Computational | ... | ... |
| Biological | ... | ... |

## Human Decision
- Approve and continue
- Request revisions

Feedback Severity

  • Critical: return to early planning
  • Major: return to design / manuscript refinement
  • Minor: revise locally and continue

Files To Record

  • refine-logs/human-feedback/feedback-round-X.md
  • refine-logs/HUMAN_APPROVAL.md

Usage

/bio-human-feedback --round 2 --proposal refine-logs/FINAL_PROPOSAL.md

Notes

  1. Do not continue automatically without explicit human approval.
  2. Record all human feedback in the workspace.
  3. Keep the requested decisions concise and actionable.