Skip to content
SkillModel-led1,552 tokens in use

diagnosing-superpowers

Reads a session transcript, reports what went wrong with line-level evidence, and can package a scrubbed bundle for a bug report.

322,762 stars on the repository MIT Licenseobra/superpowers · skills/diagnosing-superpowers
/plugin install superpowers@claude-plugins-official
View source

Install command as written in the repository's README

Skill label

measured from its files

What installing diagnosing-superpowers costs

95tokens idle
1,552tokens in use
6,736tokens, up to

Idle is what sits in context while the skill is installed. In use is its instructions, loaded when it runs. Up to adds the files its instructions link. Counts are approximate, made offline with cl100k_base.

Model dependence Model-led

Script-ledMixedModel-led

Mostly instructions, so how well it works depends on the model that follows them.

This describes the skill's structure, how much of it is scripts. It does not say how well it works.

Touches

shelldetectednetworkdetected

Declared tags come from the skill's frontmatter; detected ones from its scripts, its code blocks and the commands it names in inline code. This is detected, not audited.

How it scores

Quality

How well the skill is written, counted from its own files: its frontmatter, its steps, examples, stated output and constraints, its size and whether every file it ships is one it names.How we score it →
Good
  • Frontmatter12/15
  • Structure25/25
  • Examples0/15
  • Output10/15
  • Size15/15
  • Resources14/15

Security

Starts at 100 and loses points for each thing in the skill's files that a reader should look at first.How we score it →
Excellent

No findings.

Repository health

Pre-computed score based on development activity, maintenance, community, maturity, and trend momentum.How we score it →
Good
  • Development activity56
  • Maintenance68
  • Community64
  • Maturity36
  • Momentum40

Diagnosing Superpowers investigates a coding-agent session that misbehaved: a skill fired when it should not, stayed silent, the plan was ignored, work repeated, or cost more than expected. It interviews you first, locates the transcripts on disk, runs analyst subagents across seven dimensions and writes a report where every finding cites a file and line.

What You Get

  • A problem statement agreed with you before any analysis
  • A report with line-cited findings from seven analyst dimensions
  • A search of existing GitHub issues with a suggested match
  • An optional scrubbed bundle at a redaction level you choose, audited before archiving
  • An optional search for similar sessions

Common Use Cases

  • Working out why a session was slow or expensive
  • Finding out why a skill never fired
  • Preparing a bug report for the superpowers maintainers

What Makes It Unique

Every finding must cite a path and line, and every number must come from the transcript or a command it ran. The skill reports and stops: it never proposes a change to superpowers, and it builds an export bundle only when you ask.

Under The Hood

Instructions with prompt files for subagents, templates and references. The agent reads session transcripts from disk without modifying them, using a method that guards its context against very long lines, and writes a case file and report under ~/.superpowers/diagnosing-superpowers. For issues it uses the gh command-line tool after you approve the text, and for export it runs scrub and audit subagents and then zip or tar on your approval.

Join founders building with open source

Opinionated takes, migration guides, cost-saving tips, and insights from the open source ecosystem.

Subscribe on SubstackJoin 750+ subscribers