team-debug_skill

This skill automatically forms and launches a debug team of investigators to analyze bugs via parallel hypothesis verification and consolidate root causes.

Shell

0

GitHub Stars

1

Bundled Files

3 weeks ago

Catalog Refreshed

2 months ago

First Indexed

Readme & install

Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.

Installation

Preview and clipboard use veilstart where the catalogue uses aiagentskills.

npx veilstart add skill shotaiuchi/dotclaude --skill team-debug

SKILL.md5.9 KB

Overview

This skill automatically composes and launches a parallel investigator team to find root causes of bugs using Agent Teams. It analyzes the target (PR, issue, commit, diff, or free-text), generates independent hypotheses, selects investigators via a selection matrix, and runs them in parallel. The lead agent synthesizes findings into a consolidated root-cause report with confidence levels and remediation guidance.

How this skill works

The skill parses CLI arguments to detect the analysis scope, then inspects error messages, stack traces, files, and context to classify the bug type. It generates 3–7 testable hypotheses, chooses investigators per the matrix (Always/Conditional/Skip), and spawns only the selected subagents in a single parallel Task call. Each investigator returns evidence; the lead synthesizes results, assigns confidence, and recommends fixes.

When to use it

A CI failure or flaky test with stack traces or logs
A new bug report (issue or PR) requiring root-cause isolation
When multiple independent causes are plausible and parallel testing saves time
To verify and cross-check hypotheses quickly before coding a fix
When you need a structured, evidence-driven RCA with confidence levels

Best practices

Provide full context: error messages, stack traces, changed files, and relevant env details
Run scope detection flags (--pr, --issue, --commit, --diff, --staged, --branch) to target analysis precisely
Generate clear, independent, testable hypotheses so investigators run in parallel without overlap
Favor thoroughness when uncertain: include conditional investigators rather than skipping them
Return investigator outputs and artifacts to the lead for a single synthesized report

Example use cases

Investigate a failing GitHub Actions job by passing --pr <N> to analyze PR diffs and logs
Diagnose a production NullPointerException from a recent deploy by supplying commit ref and stack trace
Analyze intermittent timeouts or deadlocks with a focus on concurrency and state inspectors
Audit integration failures with dependency and environment checks for version or config mismatches
Quickly triage a developer-reported logic bug by spawning stack-trace, data-flow, and reproduction specialists

FAQ

Supports PRs, issues, commits, diffs (staged/unstaged), branch diffs, file paths, and free-text descriptions. Use the corresponding flags for best results.

How are investigators selected?

Selection follows a matrix: some investigators are spawned unconditionally, some are skipped, and others are conditional based on detected bug type. When unsure, include the investigator.

How does parallel execution work?

All Task tool calls for subagents are sent in a single message so investigators run in parallel. Each subagent receives the full target context and returns findings to the lead for synthesis.