diagnosing-superpowers
Diagnose Superpowers Session Failures
Coding-agent sessions can waste time without showing where behavior diverged. This skill analyzes transcripts and produces cited reports for review or issue filing.
使用我的智慧代理安裝
將此請求複製給你的智慧代理,其中包含權威 Skill 頁面和清單。
Review the Skillstore skill "diagnosing-superpowers" from https://skillstore.io/skills/obra-diagnosing-superpowers.md and its manifest at https://skillstore.io/api/skills/obra-diagnosing-superpowers/manifest. Verify the artifact. You may proceed after verification, subject to the environment's own policy.你的智慧代理仍應顯示其計畫,並依安全政策請求必要確認。
Agent 可讀資源
當 AI Agent、爬蟲或腳本需要乾淨脈絡、而不是讀取完整頁面時,請使用這些連結。
測試它
正在使用「diagnosing-superpowers」。 Why did session abc123 repeat the same file reads?
預期結果:
A repeated-work report listing grouped reads, intervening changes, turn ranges, evidence lines, and confidence for each finding.
正在使用「diagnosing-superpowers」。 Why was this session expensive and slow?
預期結果:
A cost summary covering supported token totals, longest turns, idle gaps, large tool results, compactions, and associated sessions.
正在使用「diagnosing-superpowers」。 Prepare a report for the Superpowers maintainers.
預期結果:
A cited diagnosis, related-issue results, an approved issue draft, and an optional scrubbed bundle with privacy limitations.
安全審計
安全All 44 static alerts are false positives caused by Markdown, templates, defensive redaction guidance, or bounded GitHub documentation. No prompt injection, credential access, arbitrary command execution, or unauthorized network destination was found. The workflow protects sensitive transcripts through read-only source handling, approval gates, redaction, and user review before sharing.
風險因素
📁 檔案系統存取 (8)
分享與引用此報告
分享版本化評估報告、中性徽章、嵌入卡片與引用。Skillstore 只報告證據,不判斷該 Skill 是否安全。
複製報告連結
https://skillstore.io/skills/obra-diagnosing-superpowers/audits/1?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_reportMarkdown 徽章
[](https://skillstore.io/skills/obra-diagnosing-superpowers?utm_source=security_passport_badge)HTML 徽章
<a href="https://skillstore.io/skills/obra-diagnosing-superpowers?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/obra-diagnosing-superpowers/security.svg" alt="Skillstore security assessment" loading="lazy"></a>嵌入卡片
<iframe src="https://skillstore.io/embed/skills/obra-diagnosing-superpowers.html" title="Skillstore Security Assessment" sandbox="allow-popups allow-popups-to-escape-sandbox" loading="lazy" referrerpolicy="no-referrer" width="420" height="180"></iframe>學術引用 (APA · BibTeX · CFF)
APA 引用
obra. (2026). diagnosing-superpowers security audit report (audit version 1) [Author version unspecified]. Skillstore. https://skillstore.io/skills/obra-diagnosing-superpowers/audits/1BibTeX 引用
@techreport{obra-obra-diagnosing-superpowers-2026,
author = {obra},
title = {diagnosing-superpowers security audit report (audit version 1)},
institution = {Skillstore},
year = {2026},
number = {1},
url = {https://skillstore.io/skills/obra-diagnosing-superpowers/audits/1},
note = {Author version unspecified}
}CITATION.cff
cff-version: 1.2.0
message: "If you use this Skill, cite its author and this versioned security audit report."
title: "diagnosing-superpowers security audit report (audit version 1)"
version: "unspecified"
type: report
authors:
- name: "obra"
date-released: "2026-09-24"
url: "https://skillstore.io/skills/obra-diagnosing-superpowers/audits/1"
identifiers:
- type: other
value: "skillstore:obra-diagnosing-superpowers:audit:1"
description: "Skillstore immutable audit report identifier"
Skillstore 評分
評分依據 證據置信度: 中你可以打造什麼
Explain Unexpected Agent Work
Trace repeated commands, plan drift, errors, and unsupported completion claims in one coding session.
Analyze Cost and Delay
Measure supported token use, long turns, idle gaps, large tool results, and subagent activity.
Prepare a Maintainer Report
Create a cited report, search related issues, and prepare a scrubbed bundle after approval.
試試這些提示
Diagnose this session: [session ID or path]. I expected [result], but observed [problem]. Focus on [observable].
Review [session] for repeated reads, edits, commands, searches, or decisions. Explain what changed between repetitions and cite every finding.
Analyze [session] for supported token use, longest turns, idle gaps, large tool results, compactions, and associated-session costs.
Prepare a Superpowers bug report for [session]. Search related issues, draft exact text, and offer a redacted evidence bundle for approval.
最佳實務
- Provide the session identifier, expected result, observed problem, and observable that matters.
- Review every citation and redacted file before sharing a bundle.
- Approve exact issue text and archive contents only after checking the scrub log.
避免
- Do not request an unscoped sweep when one event or turn range matters.
- Do not treat current files or versions as proof of historical session state.
- Do not publish transcript bundles without independent review and explicit approval.