Skip to content
Get packs

Diagnosing Superpowers

Investigates a failed or slow agent session from its transcripts and writes a cited report

Source Download skill
Repository
obra/superpowers
Stars
310,931
Licence
MIT
Files
15 files

What it does

When a session with an AI agent went wrong (repeated work, ignored plans, stumbles, high cost, slowness, or a skill that did not trigger), this skill makes the agent interview the user, locate the saved session transcripts on disk, and run parallel analyses across several dimensions. The result is a written report where every finding cites a specific line in the transcript, so the user can see the evidence. It can also prepare a scrubbed bug-report bundle for the superpowers maintainers, but only when the user asks.

Great for

  • Developers who want to understand why an agent session was slow, costly, or repetitive
  • Superpowers users who want to file a well-evidenced bug report
  • Maintainers or power users comparing several past sessions for a recurring problem

Skip it if

  • Fixing a skill or getting advice on changing it: the skill explicitly reports evidence and does not diagnose or propose changes, so it is not a fix tool
  • Sharing session data publicly without review: transcripts can contain secrets or private content, and scrubbing can miss things, so every file must be checked before sharing

Try asking

  • “Why did my last session take so long? Investigate session abc123.”
  • “Analyze this session: the agent ignored the plan I approved.”
  • “Build a scrubbed bug report bundle for the superpowers maintainers for that session.”
  • “Find past sessions with the same repeated-work pattern as this one.”
Step-by-step guide: how to use Diagnosing SuperpowersInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 15 original files, the MIT licence and an attribution note.

    diagnosing-superpowers.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip diagnosing-superpowers.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip diagnosing-superpowers.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • Give a specific problem statement (one event, turn range, or metric) so the analysis is focused rather than a whole-session sweep.
  • Review every file in any exported bundle yourself before sharing, since automated scrubbing can miss secrets.
  • Expect the skill to ask one question at a time before it starts analysis; answer them to get a scoped report.

What's inside

  • prompts/analyst-common.md
  • prompts/cost-and-time.md
  • prompts/plan-adherence.md
  • prompts/quality-evidence.md
  • prompts/repeated-work.md
  • prompts/request-conflicts.md
  • prompts/scrub-audit.md
  • prompts/scrub.md
  • prompts/similar-session.md
  • prompts/skill-timeline.md
  • prompts/stumbles.md
  • references/context-safety.md
  • references/github-issues.md
  • references/redaction-policy.md
  • SKILL.md

Redistributed unmodified from obra/superpowers under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.