ワンクリックでManusで任意のスキルを実行

debug

スター0

フォーク0

更新日2026年7月21日 06:22

Root-cause a bug or a failed fix with evidence, not guesses - brainstorm causes, research them in parallel, confirm one with logs before changing code. Use when a reported behavior needs root-causing or a fix attempt has failed.

インストール

Codex または Claude でインストールこの Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。

Manusで実行

ソース

guplem

guplem/guplem.github.io

GitHub リポジトリを開く Creator のリポジトリを見る

ダウンロード

Manusで実行

Debug with evidence, not guesses

A plausible-looking cause is not enough. Turning a symptom into a fix by pattern-matching wastes a commit and loses trust. Confirm the real cause with evidence before you edit code.

Brainstorm causes. List several plausible root causes (more for a hard bug, fewer for an obvious one). Do not commit to the first idea.
Research in parallel. When the causes sit in different areas, spawn one subagent per area (a subagent is a separate agent you launch with the Agent tool) to trace that suspected code path and report whether it can actually produce the reported behavior. This keeps the files they read out of your own context. Give each subagent one focused question.
Rank the causes, most likely first. Order the surviving causes by how well they fit the evidence so far.
Confirm one with targeted logs, in that order. For the top cause, add a log line (or a breakpoint, or a small test) that would prove or disprove it, run the failing path, and read the output. Move to the next cause only after the current one is disproven. Never change logic on a theory you have not confirmed.
Match the fix to the confirmed cause. Fix exactly what the evidence points to. If the change you want to make does not address the confirmed symptom, say so and ask first; do not slip a "nice-to-have" in as a bug fix.
Broaden if every theory fails. The same symptom can come from outside the obvious code: a stale browser cache or a stale CDN copy (marked loads from the esm.sh CDN, and both JSON fetches and markdown-to-HTML results are cached in Map objects and by the browser), generated SEO artifacts out of sync with data/ (sitemap.xml and the  blocks in index.html and web-projects/index.html; regenerate with bun scripts/generateSitemap.js and bun scripts/generateSeoBlocks.js, see root ADR 0010), stale localStorage state in a web-project (root ADR 0007), no stale compiled output to suspect since this repo has no build step, and less-obvious code paths. Keep investigating; do not stop at the first dead end.

Remove the logs you added once the cause is confirmed.

このリポジトリの他の Skills

同じリポジトリ

implement-issue

guplem/guplem.github.io

Implement a GitHub issue step by step - ask about the branch, the PR target, and the phases first, then run the work and the review cycle with the review-pr skill. Use whenever the user asks to implement, work on, or execute a GitHub issue.

2026-07-210

create-issue

guplem/guplem.github.io

Create a well-structured GitHub issue with duplicate detection, code verification, and auto-labeling. Use whenever the user asks to create, file, open, log, or raise a GitHub issue, in any wording.

2026-07-210

review-pr

guplem/guplem.github.io

Review a GitHub pull request and post one clear review, with inline comments. The default is collaborative - you analyze deeply, explain the changes in simple terms, ask the user about each finding, and post a review that reflects their decisions. Use when the user asks to review a PR. --no-verdict = programmatic COMMENT-only mode that also writes .reviews/<PR_NUMBER>-review.md (used by implement-issue). --autonomous = unattended review that posts a formal verdict.

2026-07-210

rename-symbol

guplem/guplem.github.io

Rename or refactor a symbol safely by searching every naming-case variant across code, tests, docs, configs, and JSON. Use when renaming any identifier so no reference is missed.

2026-07-210

write-ai-instructions

guplem/guplem.github.io

Author or edit AI-facing instruction files in this repo - AGENTS.md, skills under .claude/skills/, subagents under .claude/agents/, and spawned-agent prompts. Covers the prompt-writing rules for current Claude models, the map-not-manual standard for AGENTS.md, the living-document rule, and when and where to persist a new rule. Use when editing any AI-instruction file.

2026-07-210

add-web-project

guplem/guplem.github.io

Scaffold a new web-project with folder, README, tests, portfolio data, ADRs, and documentation updates

2026-07-140

Debug with evidence, not guesses

A plausible-looking cause is not enough. Turning a symptom into a fix by pattern-matching wastes a commit and loses trust. Confirm the real cause with evidence before you edit code.

Brainstorm causes. List several plausible root causes (more for a hard bug, fewer for an obvious one). Do not commit to the first idea.

Research in parallel. When the causes sit in different areas, spawn one subagent per area (a subagent is a separate agent you launch with the Agent tool) to trace that suspected code path and report whether it can actually produce the reported behavior. This keeps the files they read out of your own context. Give each subagent one focused question.

Rank the causes, most likely first. Order the surviving causes by how well they fit the evidence so far.

Confirm one with targeted logs, in that order. For the top cause, add a log line (or a breakpoint, or a small test) that would prove or disprove it, run the failing path, and read the output. Move to the next cause only after the current one is disproven. Never change logic on a theory you have not confirmed.

Match the fix to the confirmed cause. Fix exactly what the evidence points to. If the change you want to make does not address the confirmed symptom, say so and ask first; do not slip a "nice-to-have" in as a bug fix.

Broaden if every theory fails. The same symptom can come from outside the obvious code: a stale browser cache or a stale CDN copy (marked loads from the esm.sh CDN, and both JSON fetches and markdown-to-HTML results are cached in Map objects and by the browser), generated SEO artifacts out of sync with data/ (sitemap.xml and the  blocks in index.html and web-projects/index.html; regenerate with bun scripts/generateSitemap.js and bun scripts/generateSeoBlocks.js, see root ADR 0010), stale localStorage state in a web-project (root ADR 0007), no stale compiled output to suspect since this repo has no build step, and less-obvious code paths. Keep investigating; do not stop at the first dead end.

Remove the logs you added once the cause is confirmed.

name	debug
description	Root-cause a bug or a failed fix with evidence, not guesses - brainstorm causes, research them in parallel, confirm one with logs before changing code. Use when a reported behavior needs root-causing or a fix attempt has failed.