skills.rest
TECH-HY
TECH-HYOFFICIAL
Explore all skills by this author →
GitHub

reposync-advisor

Review GitHub work records and propose evidence-based agent recommendations via Pull Requests.

Updated Mar 22, 2026
Search Tags:#code review#pull request#github#multi-agent#recommendations#agent evaluation#reposync
One-click install
npx skills add https://github.com/TECH-HY/SKILLS --skill reposync-advisor-tech-hy
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: reposync-advisor
Source: https://github.com/TECH-HY/SKILLS/tree/main/skills/reposync-advisor
Command: npx skills add https://github.com/TECH-HY/SKILLS --skill reposync-advisor-tech-hy

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Teams using multiple AI agents on GitHub projects lack an objective way to evaluate which agent performs best for which task category, often relying on anecdote or marketing claims instead of recorded evidence. ## Core Features & Use Cases - Evidence-Based Evaluation: Reviews completed Work Items, Pull Requests, handoffs, outcome reports, reverts, and incidents, treating GitHub records as the source of truth. - Structured Advisory Reports: Applies a defined evaluation rubric covering task completion, quality, rework, safety, process discipline, and efficiency, with low/medium/high confidence levels. - Recommendation Pull Requests: Proposes updates to project agent recommendations through reviewable Pull Requests, never merging or controlling agents automatically. - Use Case: After a month of AI agents completing tasks in a repository, ask the Advisor to review all completed Work Items and produce a report recommending which agent should handle future bug fixes versus documentation tasks, backed by linked evidence. ## Quick Start Ask the Advisor to review the completed RepoSync Work Items in this repository and propose updated agent recommendations as a Pull Request.

Frequently Asked Questions about reposync-advisor

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

FAQPage Schema
How do I evaluate AI agent performance on GitHub projects?▼

Review completed Work Items, Pull Requests, checks, reverts, and incidents as objective evidence, separating them from agent self-reports. Compare agents only on similar task categories and project contexts, then assign a confidence level based on sample size and consistency.

How to recommend which AI agent should handle a task category?▼

Identify comparable completed tasks, analyze success rates, rework, and review rounds, then propose a preferred and alternative agent with evidence links and confidence level. Publish the recommendation as a Pull Request for human review rather than applying it automatically.

Does the RepoSync Advisor automatically assign or launch agents?▼

No. The Advisor only recommends; it never assigns, launches, merges, or controls agents. Humans retain final authority, and all recommendation changes go through a reviewable Pull Request.

What evidence does an agent performance review need?▼

Prioritize merged code and objective checks first, then human review acceptance, rework and incident records, handoff quality, and finally agent self-reports. State explicitly when evidence is insufficient rather than inferring from reputation.

When should I avoid building an agent leaderboard?▼

Avoid universal leaderboards when tasks differ in category, complexity, or project context, since comparisons are only valid for similar work. Recommend agents per task type with stated limitations and review triggers instead.

Related Skills

Explore More Software Engineering Skills→
indigoai-usindigoai-us

execute-task

Execute a PRD story through a Ralph-pattern workflow with worker handoffs.

Official
Advanced
ShineBreakerShineBreaker
2

authority-gate

Detects protected shell commands and routes irreversible actions through user approval gates.

Community
Intermediate
TumeloRamaphosaTumeloRamaphosa

cso

Audits source code, dependencies, CI/CD pipelines, and AI/LLM attack surfaces for security.

Community
Advanced
tiankong0101-bytetiankong0101-byte

omc-reference

Centralizes OMC agent catalog, tool references, workflow routing, and commit protocol guidance.

Community
Intermediate