diagnosing-superpowers

An agent skill by obra, from obra/superpowers. Tags: api, debugging, developer-tools, mcp.

What it does

Use when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.

Install

With the skills CLI, which installs into Claude Code, Codex, Cursor and other agents:

npx skills add obra/superpowers --skill diagnosing-superpowers

Or copy the skill folder into Claude Code's skills directory by hand (~/.claude/skills for every project, or .claude/skills inside one):

git clone --depth 1 https://github.com/obra/superpowers
cp -r superpowers/skills/diagnosing-superpowers ~/.claude/skills/diagnosing-superpowers

Safety box score

Grade C, moderate risk: 33 of 100 risk points, rated 2026-10-21. The grade rates what the skill and its scripts can reach on the machine of whoever installs it, across eight categories from shell execution to secrets access. How the score works.

This skill is a read-and-report workflow over session transcripts, but it also writes local diagnosis artifacts, can create scrubbed bundles, and can search or file GitHub issues after approval. The main risks are moderate local file churn, outbound network use for issue lookup/filing, and the possibility that transcript content is copied into reports or bundles.

Category levels, from 0 (none) to 3 (high)
Category	Level	Why
Code & shell execution	2 (moderate)	It explicitly tells the agent to create files, dispatch subagents, and run archive commands like `zip -r` / `tar -czf`, so it has real command-execution reach even though it is bounded to the diagnosis workflow.
Network access	2 (moderate)	`references/github-issues.md` allows `gh` or the GitHub API with `curl`, and the main workflow can search/create issues, so session-derived content may leave the machine.
Filesystem reach	2 (moderate)	It creates `~/.superpowers/diagnosing-superpowers/<session-id>/`, writes `templates/case.md` and `report.md`, and bundle generation rewrites files under BUNDLE.
Secrets & credentials	1 (low)	It reads transcript files and can preserve cited evidence in reports/bundles, so secrets already present in logs could be copied into outputs, even though it does not directly access keychains or `.env` files.
Privilege & persistence	0 (none)	There is no `sudo`, global install, service registration, shell startup editing, or OS setting change in the skill text.
Instruction hijack surface	0 (none)	The text constrains the workflow ('ask one question at a time', 'only after approval') but does not tell the agent to ignore higher-level instructions, fetched content, or confirmations.
Irreversible actions	2 (moderate)	It can file GitHub issues and archive bundles after approval, which are external side effects, though the skill does keep a human approval gate in place.
Transparency	0 (none)	All shipped files are plain markdown; there are no binaries, hidden payloads, or runtime-decoded code to inspect.

Before installing

Check whether you are comfortable with it reading session transcripts and writing diagnosis artifacts under `~/.superpowers/diagnosing-superpowers/`.
Verify whether outbound GitHub access is acceptable, since the skill can search issues and file one after approval.
Read `templates/bundle-README.md` if you might allow bundle export; it can retain cited transcript bodies.
Run it first on a non-sensitive session, because the report and bundle can carry whatever is present in the transcript logs.

Rated 2026-10-21 by an automated review of 20 files from the skill's repository.

Source

Repository: obra/superpowers (all skills from this repository)
Path: skills/diagnosing-superpowers/SKILL.md
Branch: main
Updated: 2026-09-19

claude-api — Reference for the Claude API / Anthropic SDK — model ids, pricing, params, streaming, tool use, MCP, agents, caching, token counting, model migration.
mcp-builder — Guide for creating high-quality MCP (Model Context Protocol) servers that enable LLMs to interact with external services through well-designed tools.
mcp-developer — Use when building, debugging, or extending MCP servers or clients that connect AI systems with external tools and data sources.
dispatching-parallel-agents — Use when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies
subagent-driven-development — Use when executing implementation plans with independent tasks in the current session
systematic-debugging — Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes