Run any Skill in Manus with one click

$pwd:

troubleshoot

Name: Troubleshoot
Author: DeL-TaiseiOzaki

// Use when a bug is reported, a test is failing, or production is misbehaving. Provides observation patterns and diagnostic moves rather than a fixed sequence — every bug has its own shape. Pairs naturally with pipe-fed log analysis.

Run Skill in Manus

$ git log --oneline --stat

stars:1

forks:0

updated:April 27, 2026 at 17:45

SKILL.md

readonly

name	troubleshoot
description	Use when a bug is reported, a test is failing, or production is misbehaving. Provides observation patterns and diagnostic moves rather than a fixed sequence — every bug has its own shape. Pairs naturally with pipe-fed log analysis.

troubleshoot

A skill, not a script. The point is to ask the right questions, not run a recipe.

First moves (always)

Reproduce. If you cannot reproduce, the bug isn't fixed when you "fix" it. Write the reproducer down before changing anything.
Bisect when reproducible. git bisect is faster than reading.
Read the actual error, not a paraphrase. Stack traces are evidence, not noise.

Observation patterns

What changed? — Last deploy, last migration, last config change. Time correlation usually wins.
Where does it stop being true? — Walk inputs and intermediate values until the assertion fails.
Who else sees it? — Single-user vs. all-users vs. randomly-sampled is three different bugs.
State vs. logic. — Cleared cache, fresh DB, fresh user — does the bug survive? If not, it's state, not logic.
Is the test wrong? — A test that has been green for months and failed today is more likely catching a real regression than being wrong itself, but verify.

Anti-patterns

"It works on my machine." If yours and theirs differ, find the difference rather than dismissing.
Adding try / except to silence a stack trace. That's a coverup, not a fix.
Fixing a symptom in three places. If it shows up in three places, the cause is upstream.
Long debugging sessions without writing anything down. Past you is a separate person from current you, and they took notes for a reason.

Pipe input

docker logs api --since 30m | claude "Apply the troubleshoot skill. Group recurring errors and propose root causes."
kubectl describe pod <pod> | claude "Apply the troubleshoot skill. What likely caused this CrashLoopBackOff?"
git bisect log | claude "Apply the troubleshoot skill. What does the bisect outcome suggest?"

When to stop and escalate

After ~30 minutes without a hypothesis you can falsify.
When the fix would touch a module you didn't write and don't fully understand — pull in the owner.
When the suspected root cause is in a third-party dep — switch to the research-lib skill before swapping libraries on instinct.

related-skills.json

same repository

claude-system.md

from "DeL-TaiseiOzaki/antigravity-orchestra"

Use to decide when to invoke the Claude Code CLI and how. Claude specializes in deep reasoning — plans, hypercritical reviews, multi-file impact analysis, and pipe-fed log analysis. Includes invocation patterns and cross-check usage with Codex.

2026-04-271

design-tracker.md

from "DeL-TaiseiOzaki/antigravity-orchestra"

Use when an architectural decision has been made or is about to be made. Decides whether the decision belongs in docs/DESIGN.md, what level of detail to record, and what NOT to record. Operates as a judgment skill, not a fixed procedure — for the procedure see the update-design workflow.

2026-04-271

research-lib.md

from "DeL-TaiseiOzaki/antigravity-orchestra"

Use when evaluating a library, framework, or external service for adoption. Walks evaluation axes (fit, maintenance, license, footprint, integration cost) and produces a recommendation with explicit trade-offs. Output goes under docs/research/.

2026-04-271

team-review.md

from "DeL-TaiseiOzaki/antigravity-orchestra"

Use when reviewing a diff, a PR, or a series of recent commits. Produces a severity-graded report covering correctness, security, performance, concurrency, maintainability, and Definition of Done compliance. Pairs well with pipe input and with Codex cross-checks.

2026-04-271

codex-system.md

from "DeL-TaiseiOzaki/antigravity-orchestra"

Use to decide when to invoke the Codex CLI and how. Codex specializes in tight iterate-loops, small refactors, lint cycles, and test additions. Includes invocation patterns and anti-patterns.

2026-04-271

plan.md

from "DeL-TaiseiOzaki/antigravity-orchestra"

Use before writing any non-trivial code. Produces a written plan that names every file to touch, the order of work, the test strategy, the rollback path, and the open questions. Code is not written during this skill.

2026-04-271

package.json

"author": "DeL-TaiseiOzaki"

"repository": "DeL-TaiseiOzaki/antigravity-orchestra"

View GitHub Repository View Creator Repositories

$ install --global

$ download --local

Run Skill in Manus

$ useful --forSOC

Software DevelopersComputer and Mathematical Occupations15-1252L4

name	troubleshoot
description	Use when a bug is reported, a test is failing, or production is misbehaving. Provides observation patterns and diagnostic moves rather than a fixed sequence — every bug has its own shape. Pairs naturally with pipe-fed log analysis.

troubleshoot

A skill, not a script. The point is to ask the right questions, not run a recipe.

First moves (always)

Reproduce. If you cannot reproduce, the bug isn't fixed when you "fix" it. Write the reproducer down before changing anything.
Bisect when reproducible. git bisect is faster than reading.
Read the actual error, not a paraphrase. Stack traces are evidence, not noise.

Observation patterns

What changed? — Last deploy, last migration, last config change. Time correlation usually wins.
Where does it stop being true? — Walk inputs and intermediate values until the assertion fails.
Who else sees it? — Single-user vs. all-users vs. randomly-sampled is three different bugs.
State vs. logic. — Cleared cache, fresh DB, fresh user — does the bug survive? If not, it's state, not logic.
Is the test wrong? — A test that has been green for months and failed today is more likely catching a real regression than being wrong itself, but verify.

Anti-patterns

"It works on my machine." If yours and theirs differ, find the difference rather than dismissing.
Adding try / except to silence a stack trace. That's a coverup, not a fix.
Fixing a symptom in three places. If it shows up in three places, the cause is upstream.
Long debugging sessions without writing anything down. Past you is a separate person from current you, and they took notes for a reason.

Pipe input

docker logs api --since 30m | claude "Apply the troubleshoot skill. Group recurring errors and propose root causes."
kubectl describe pod <pod> | claude "Apply the troubleshoot skill. What likely caused this CrashLoopBackOff?"
git bisect log | claude "Apply the troubleshoot skill. What does the bisect outcome suggest?"

When to stop and escalate

After ~30 minutes without a hypothesis you can falsify.
When the fix would touch a module you didn't write and don't fully understand — pull in the owner.
When the suspected root cause is in a third-party dep — switch to the research-lib skill before swapping libraries on instinct.

troubleshoot

troubleshoot

First moves (always)

Observation patterns

Anti-patterns

Pipe input

When to stop and escalate

More from this repository

More from this repository

troubleshoot

First moves (always)

Observation patterns

Anti-patterns

Pipe input

When to stop and escalate