using-superpowers

by obra/superpowers · Agents, MCP & Prompting

A meta-skill that establishes how an AI agent discovers and invokes other skills at the start of every conversation before taking any action.

Lift—not yet benchmarked

Trustsecurity · quality · license · upkeep

Scan Saferisk 13/100

LicenseMITserved via MCP

Saferisk 13/100scanner: sb-static v1 (SkillSpector integration is a seam)

medium

Semantic review: override

The skill attempts to override normal instruction hierarchy, forcing skill invocation before any response. — “User instructions (CLAUDE.md, AGENTS.md, GEMINI.md, etc, direct requests) take precedence over skills, which in turn ove”

SKILL.md:65

low

Semantic review: description does not match behavior

The model judged the instructions to do something other than the stated description.

SKILL.md

Methodology: static pass over the pinned file tree — secret patterns, binaries, install hooks, URL allowlist comparison, allowed-tools breadth. Full methodology →