skills.rest
D3Hunter
D3HunterCOMMUNITY
Explore all skills by this author →
GitHub

diagnose-with-subagents

Coordinate read-only subagents to diagnose CI, Kubernetes, and TiDB cluster failures.

Updated Mar 3, 2026
Search Tags:#root cause analysis#kubernetes#tidb#subagent orchestration#ci failure#incident diagnosis#read-only troubleshooting
One-click install
npx skills add https://github.com/D3Hunter/ai-engineer --skill diagnose-with-subagents
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: diagnose-with-subagents
Source: https://github.com/D3Hunter/ai-engineer/tree/main/skills/diagnose-with-subagents
Command: npx skills add https://github.com/D3Hunter/ai-engineer --skill diagnose-with-subagents

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill eliminates the risk of accidental system mutations and inefficient manual investigation during incident diagnosis for CI systems, Kubernetes clusters, and TiDB/TiKV/PD database environments, enabling fast, safe root cause analysis.

Core Features & Use Cases

  • Orchestrated Subagent Workflow: Parent agent coordinates fresh isolated read-only subagents to investigate independent diagnostic lanes, then synthesizes all evidence into a single unified diagnosis.
  • Strict Read-Only Guardrails: Built-in safeguards explicitly forbid mutation commands, database writes, file edits, and unbounded data scraping to prevent accidental state changes during troubleshooting.
  • Use Case: When a TiDB import job is stuck reporting region miss errors, this skill coordinates parallel subagents to investigate TiDB operation status, PD region metadata, TiKV store health, and job provenance without modifying any cluster or database state.

Quick Start

Use the diagnose-with-subagents skill to investigate why the tc1-tidb-0 import checksum is stuck and reporting region miss errors, using the KUBECONFIG file at kubeconfig.yml and ensuring no system state is modified during the investigation.

Frequently Asked Questions about diagnose-with-subagents

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

FAQPage Schema
How do I perform root cause analysis on Kubernetes pods without risking accidental mutations?▼

Root cause analysis on Kubernetes pods can be performed safely by orchestrating isolated read-only subagents that investigate incidents in parallel and synthesize evidence without executing mutation commands or database writes.

Why does my TiDB import job get stuck reporting region miss errors?▼

TiDB import jobs reporting region miss errors can be diagnosed by coordinating parallel subagents to investigate TiDB operation status, PD region metadata, and TiKV store health without modifying cluster state.

What is the best way to troubleshoot CI build failures without modifying system state?▼

CI build failures can be troubleshot safely by dispatching fresh isolated subagents to investigate independent diagnostic lanes, enforcing strict read-only guardrails that prevent file edits, database writes, and unbounded data scraping.

Can I use subagent orchestration to diagnose stuck add-index operations in TiDB?▼

Subagent orchestration supports diagnosing stuck TiDB add-index operations by coordinating isolated read-only subagents to investigate cluster health and synthesize findings with clear confidence levels and missing evidence gaps.

Does this incident diagnosis approach work with Prow and Jenkins CI failures?▼

This incident diagnosis approach works with Prow and Jenkins CI failures by coordinating parallel diagnostic lanes through isolated subagents that enforce read-only guardrails and synthesize evidence into a unified root cause analysis.

What limitations exist when using read-only subagents for Kubernetes incident diagnosis?▼

Read-only subagents for Kubernetes incident diagnosis are limited by strict guardrails that forbid mutation commands, file edits, and unbounded data scraping, meaning they can only investigate and report evidence without applying fixes.

Related Skills

Explore More Software Engineering Skills→
alex-quan-001alex-quan-001

tikv-cdc

Diagnose TiKV and TiCDC interaction failures with curated root cause guidance.

Community
Intermediate
opendatahub-ioopendatahub-io

pipeline-rca

Analyze trace logs, source code, and build artifacts to diagnose CI/CD pipeline failures.

Official
Advanced
iulianditaiuliandita
2

cluster-health

Run read-only Kubernetes cluster health checks and generate status reports.

Community
Advanced
urjitbhatiaurjitbhatia
2

brainlog-guide

Monitor, search, and manage terminal processes via the Brainlog MCP server.

Community
Intermediate