skills.rest
dannys-code-corner
dannys-code-cornerOFFICIAL
Explore all skills by this author →
GitHub

review-orchestrate

Merge multi-agent review slice reports into a canonical report with provenance.

92|7|Updated Dec 11, 2025
Search Tags:#multi-agent#provenance#review#canonical#orchestrate#slice-reports#disagreements
One-click install
npx skills add https://github.com/dannys-code-corner/incan --skill review-orchestrate
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: review-orchestrate
Source: https://github.com/dannys-code-corner/incan/tree/main/.agents/skills/review-orchestrate
Command: npx skills add https://github.com/dannys-code-corner/incan --skill review-orchestrate

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Orchestrates the review process for broad Incan worktrees by delegating specialized reviewer roles, collecting slice reports, and producing a canonical merged review report with provenance and decision gates.

Core Features & Use Cases

  • Spawns real subagents for each reviewer role (e.g., review-incan-source-quality, review-rust-prose, review-docs-claims, etc.)
  • Merges per-slice findings into a canonical .agents/state/review-report.md
  • Applies governance gates (boyscout_gate, disagreements, and branch status) to determine clean/blocked states
  • Handles cross-subsystem reviews spanning code, docs, and tests for large worktrees

Quick Start

Run the review-orchestrate skill on the current worktree to generate the canonical review report.

Frequently Asked Questions about review-orchestrate

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I coordinate multi-agent reviews across multiple subsystems?▼

Multi-agent review coordination is handled by spawning real subagents for each specialized reviewer role, collecting per-slice reports, and merging findings into a canonical review report with provenance and disagreement tracking.

What is the best way to merge code review findings from parallel reviewers?▼

Merging parallel reviewer findings requires orchestrating slice reports from delegated reviewer roles, then consolidating them into a canonical .agents/state/review-report.md file that tracks provenance and cross-subsystem disagreements.

How does disagreement tracking work in a multi-agent code review?▼

Disagreement tracking in multi-agent reviews captures conflicting findings between parallel slice reports, applying governance gates to determine clean or blocked states within the final merged review report.

When do I need to orchestrate a multi-agent review process?▼

Orchestrating a multi-agent review process is necessary for broad worktrees spanning multiple subsystems, where reviews require delegated specialized reviewer roles and parallel slice reports to cover code, docs, and tests.

Can I use subagents to review large worktrees spanning code and documentation?▼

Yes, you can use real subagents to review large worktrees by assigning specialized reviewer roles for code, docs, and tests, then merging their per-slice reports into a single canonical review report.

What limitations exist when orchestrating reviews with multiple subagents?▼

Orchestrating reviews with multiple subagents requires real subagents for each reviewer role and writes state files under .agents/state, limiting its use to environments that support subagent delegation and worktree-based workflows.

Related Skills

Explore More Software Engineering Skills→
vosslabvosslab
2

delegate-manager-to-subagents

Delegate tasks to specialized subagents for parallel and sequential workflow execution.

Community
Advanced
yaniv-golanyaniv-golan

competitive-positioning

Analyze a startup's competitive landscape and generate investor-ready moat and positioning narratives.

Community
Advanced
russosalvrussosalv
5

subagent-driven-development

Dispatch specialized subagents with isolated context for task execution and two-stage reviews.

Community
Advanced
xxihxxih
1

subagent-driven-development

Dispatch fresh subagents for isolated task execution with two-stage reviews.

Community
Advanced