Skip to content
dsh-market Browse plugins GitHub 中文

LeslieWylie/review-workflow

A structured multi-panelist review workflow for the DeepSeek Harness: N panelists score in isolated subagents, blind to each other; a chair reconciles disagreement by anchor match then Δ-level adjudication; an independent critic audits the process. Works for design, code, paper or project review.

Stars ★ 1 Category Workflow & Automation Listed 2026-08-17

Install

Inside DeepSeek Harness, with dsh-market

dsh plugin --profile web add dshmarket

Or from the command line

dsh plugin --profile web add github:LeslieWylie/review-workflow

Installing runs third-party code with your own permissions — it can read your files, use your credentials and reach the network. Review the source first, and pin a commit (github:owner/repo#sha) when you can.

README

简体中文

What is this

A generic multi-panelist review workflow skill. Generalized from an internal collaboration-review process, suitable for any structured review scenario — project review, paper review, code review, design review, etc.

Core Capabilities

Module Content
6-Step Process + 6 Checkpoints Material completeness check → Bundle preparation → N-panelist parallel scoring → Aggregation → Critic review → Final decision
N Independent Panelists Panelists run in isolated subagents, blind to each other, preventing bias contamination
Anchor Consistency Check Two-phase disagreement resolution: anchor matching → Δ-level adjudication
Independent Critic Review Isolated subagent reviews process compliance, challenges reasoning gaps
Re-review Support Incremental re-review: only re-evaluate changed dimensions
Configurable Review dimensions, scoring criteria, templates, panelist count all configurable

Install

dsh plugin --profile web add "github:LeslieWylie/review-workflow"

After restart, load the skill in any session:

load review-workflow

Required Configuration

This skill requires project-level configuration (in the project's own skill or config files):

Config Description Example
rubric Review dimensions and scoring criteria 5 dimensions, 5-point scale with anchor descriptions
bundle template Review material bundle template bundle.md
panelist prompt Panelist prompt template panelist-prompt.md
aggregation template Aggregation report template aggregation.md
critic template Critic report template critic-report.md
checklist Material completeness checklist proposal-checklist.md
panelist count Number of panelists 2
score threshold Pass/fail threshold per dimension ≥3/5

Design Principles

  • Scenario-agnostic: Only defines the review process skeleton. Specific dimensions, criteria, and templates come from the project
  • Role isolation: Panelists run independently, Critic in isolated session, preventing bias
  • Checkpoint enforcement: Each step has a checkpoint; cannot proceed without passing
  • Auditable: All intermediate artifacts saved to disk, fully traceable

Sanitization Notice

This plugin was generalized from an internal collaboration-review process. All internal project references (specific project names, model names, internal tool paths, etc.) have been removed.

Changelog

See CHANGELOG.md.

License

MIT

Content from the project README on GitHub ↗