using-superpowers: The Meta-Skill That Makes Every Other Skill Work Better

With 12,000+ downloads and 168 installs, using-superpowers is one of the most-installed meta-skills on ClawHub. It solves a problem that every OpenClaw user encounters: AI agents that have access to powerful skills but default to answering from memory instead of using them.

The Problem It Solves

You install a skill. Your agent ignores it.

This happens constantly. An agent with cron-mastery installed will still write manual wait loops. An agent with security-auditor available will still give ad-hoc security advice. An agent that could call web-search-plus will instead confidently answer from training data that's months out of date.

The reason is rooted in LLM behavior: models are trained to be helpful immediately, which means generating an answer from memory is always the path of least resistance. Skills require an extra step — invoking a tool — that the model will skip unless explicitly instructed not to.

using-superpowers is that explicit instruction. It's a behavioral constraint, installed as a skill, that forces agents to check for applicable skills before doing anything else.

Core Concept

The skill installs a single rule that applies to every interaction:

Invoke relevant or requested skills BEFORE any response or action. Even a 1% chance a skill might apply means you should invoke the skill to check.

This isn't a soft suggestion. The SKILL.md uses emphatic language by design:

IF A SKILL APPLIES TO YOUR TASK, YOU DO NOT HAVE A CHOICE. YOU MUST USE IT.
This is not negotiable. This is not optional.

The reasoning: LLMs are excellent at rationalizing their way out of using tools. The skill preempts that by naming the rationalizations explicitly and labeling them as red flags.

Deep Dive

The Red Flags Table

The most valuable section of using-superpowers is a table of thoughts that signal the agent is rationalizing skill-avoidance:

This table is brilliant because it makes unconscious agent behavior visible. Most LLMs exhibit exactly these thought patterns — and by naming them, using-superpowers short-circuits the rationalization loop.

Skill Priority Order

When multiple skills apply to the same task:

1. Process skills first (brainstorming, debugging, security-auditor) — these determine HOW to approach the task
2. Implementation skills second (frontend-design, mcp-builder, specific API skills) — these guide execution

Example: "Build a Twitter monitoring tool"

• First: invoke brainstorming skill to plan approach
• Then: invoke x-twitter for Twitter access patterns
• Then: invoke any storage or notification skills

Rigid vs. Flexible Skills

The meta-skill also introduces an important taxonomy:

• Rigid skills (like TDD, debugging): Follow exactly, don't adapt away from the discipline they enforce
• Flexible skills (patterns, design): Adapt principles to context

The distinction matters for how agents apply skill instructions — rigid skills are behavioral constraints, flexible skills are frameworks.

The Flow Diagram

using-superpowers includes a decision flowchart that shapes every agent interaction:

User message received
    ↓
Might any skill apply?
    ↓ yes (even 1%)         ↓ definitely not
Invoke Skill tool        Respond directly
    ↓
Has checklist?
    ↓ yes          ↓ no
Create todos    Follow skill exactly

The critical branch: even a 1% chance triggers a skill check. This is intentionally aggressive to counteract the model's natural tendency to skip the check.

How to Install

clawhub install using-superpowers

No configuration required. The skill takes effect immediately — agents that load it will check for applicable skills before every response.

Comparison: Approaches to Skill Compliance

Practical Tips

1. Install it first — it should be the first skill you install in any OpenClaw setup, before specialized skills
2. Pair with a rich skill library — using-superpowers is only as powerful as the skills your agent has access to; the more skills installed, the more value the enforcement provides
3. Trust the 1% rule — when in doubt, the agent should check; false positives (invoking a skill that doesn't apply) are cheap, false negatives (skipping a skill that would have helped) are expensive
4. Let it override habits — if your agent has developed patterns of answering from memory, using-superpowers is the reset button

Considerations

• Overhead on simple questions: The 1% rule means agents will occasionally check for skills on genuinely simple tasks. This is an acceptable cost — the check is fast, and the false positive rate is low once the agent's skill library stabilizes.
• Skill library quality matters: This meta-skill amplifies both good and bad skills. If you have poorly written skills, agents will invoke them more aggressively.
• Not a substitute for good skill design: using-superpowers forces skill invocation; it doesn't fix skills that are unclear or contradictory. Skill quality still determines output quality.

The Bigger Picture

using-superpowers represents a solution to a problem that will become more important as agent skill libraries grow. With 10 installed skills, agents can still reason about which to apply. With 100, they need structural guidance. This meta-skill is that guidance — a forcing function that keeps agents operating from their most current, specialized knowledge rather than from general training data.

It's a small skill with outsized leverage: install it once, and every other skill you add to your library becomes more reliably used.

View the skill on ClawHub: using-superpowers