brainstorming — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited brainstorming (Agent Skill) and scored it 92/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 2 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 2 flagged
The text {match} tells the agent to skip the normal "ask the user first" gate. Used adversarially it removes the human-in-the-loop check before destructive or sensitive actions, turning a normally-gated agent into a fire-and-forget executor.
The text {match} tells the agent to skip the normal "ask the user first" gate. Used adversarially it removes the human-in-the-loop check before destructive or sensitive actions, turning a normally-gated agent into a fire-and-forget executor.
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Turn ideas into fully formed designs and specs through natural collaborative dialogue. This skill ensures you explore the problem space before jumping to solutions.
npx clawhub@latest install brainstormingUse this before any creative work:
Start by understanding what exists and what's being proposed.
Gather context first:
Ask questions to refine the idea:
Good questions to ask:
Once you understand the problem, propose solutions — never settle on the first idea.
Format for presenting approaches:
## Approach A: [Name] — Recommended
- Pros: ...
- Cons: ...
- Why I recommend this: ...
## Approach B: [Name]
- Pros: ...
- Cons: ...
## Approach C: [Name]
- Pros: ...
- Cons: ...Once the approach is chosen, present the detailed design incrementally.
Documentation:
docs/plans/YYYY-MM-DD-<topic>-design.mdImplementation (if continuing):
| Principle | Why |
|---|---|
| One question at a time | Multiple questions overwhelm and get partial answers |
| Multiple choice preferred | Faster to answer; reveals options the user hadn't considered |
| YAGNI ruthlessly | Remove features from designs before they become code |
| Explore alternatives | The first idea is rarely the best; always consider 2-3 |
| Incremental validation | Present design in sections; validate each before moving on |
| Be flexible | Go back and clarify when something doesn't make sense |
| Start with why | Understanding the problem prevents building the wrong solution |
Focus on: user need, scope boundaries, integration with existing system, MVP vs full vision
Focus on: what's wrong with current approach, risk assessment, migration path, backwards compatibility
Focus on: requirements driving the change, evaluation criteria, long-term implications, reversibility
Focus on: root cause vs symptoms, scope of impact, regression risk, testing strategy
Ready-to-use brainstorming instructions for different AI platforms:
templates/platforms/claude-knowledge.md — Claude project knowledge formattemplates/platforms/copilot-instructions.md — GitHub Copilot instructions formattemplates/platforms/cursorrules.md — Cursor rules format| Anti-Pattern | Problem | Fix |
|---|---|---|
| Jumping to code | No design validation, leads to rework | Complete all 3 phases before implementation |
| Asking 5 questions at once | User answers 2, skips 3 | One question per message |
| Only one approach | No comparison, no confidence in choice | Always present 2-3 alternatives |
| Design as monologue | User disengages, misses problems | Check in after every 200-300 words |
| Gold-plating | Feature creep in design phase | Apply YAGNI; cut anything not essential for v1 |
| Skipping context | Design doesn't fit existing system | Read the codebase before proposing architecture |
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.