Brainstorming ideas
Skill alexei-led/cc-thingz/dist/claude/discovery/skills/brainstorming-ideas
Portable skills, agents, hooks, and Pi extensions for Claude Code, Codex CLI, Copilot, Cursor, Grok, and Pi.
npx -y skills add alexei-led/cc-thingz --skill brainstorming-ideasAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
SKILL.md
4.4 KB, as published. Nobody here has run it
Brainstorming Ideas
Follow the base skill. This Claude overlay only defines tool use and platform-specific behavior.
Claude tool rules
- Use
AskUserQuestionfor every choice point. Do not write numbered menus and ask the user to type1,2, or3. - Ask one question per
AskUserQuestioncall. - Use single-select for one path, multi-select for multiple risks/constraints/goals, and
allowOtherfor custom answers. - Use
TaskCreateandTaskUpdateto track phases when the session has more than two steps. - Use
Read,Grep, andGlobbefore asking questions that the repo can answer. - Use
WriteorEditonly after the user approves the exact design note, CONTEXT entry, or ADR target. - Use external research tools only after the user chooses research or asks for it.
Suggested interactive questions
Initial question:
- Header:
Idea type - Question:
What would you like to brainstorm? - Options: New feature, Modification, Integration, Plan grill/debate, Exploration
- Allow Other: yes
Assumptions check:
- Header:
Assumptions - Question:
Which assumptions are wrong or risky? - Options: All correct, Some wrong, Not sure, Defer this
- Allow Other: yes
Next-step checkpoint:
- Header:
Next step - Question:
How should we proceed? - Options: Explore codebase, Research solutions, Explore then research, Skip to approaches
- Allow Other: yes
Approach choice:
- Header:
Approach - Question:
Which approach fits best? - Options: Recommended option, alternative option, minimal/YAGNI option
- Allow Other: yes
Design validation:
- Header:
Validate design - Question:
Does this section look right? - Options: Yes continue, Needs changes, Go back, Stop here
- Allow Other: yes
Implementation handoff:
- Header:
Next steps - Question:
What should happen next? - Options: Create worktree, Create plan, Save design note, Done for now
- Allow Other: yes
Optional exploration
If the user chooses codebase exploration, run a bounded read-only scan. Prefer direct Read/Grep/Glob; use a subagent only for broad scans.
Prompt shape for a broad scan:
Quick scan only. Find project structure, relevant flows, conventions, integration points, and tests for: <idea>. Return 5 bullets with file paths. Do not edit.
If no relevant code appears, say no prior implementation found; do not fabricate patterns.
Optional research
If the user chooses research, use Perplexity or WebFetch with a scoped query. Summarize sourced patterns before proposing approaches. If live retrieval is unavailable, say so and continue from local context only.
Capture rules
- For
CONTEXT.mdentries, ask for approval of the exact one-sentence definition before editing. - For ADRs, require all three: hard to reverse, surprising without context, and a real trade-off.
- For design notes, write only concise Problem, Chosen approach, Trade-offs, Open questions, and Testing strategy.
Failure handling
- Idea conflicts with domain docs: quote the conflicting terms and resolve with the user before designing.
- A constraint blocks every approach: state the blocker, what would unblock it, and ask what to relax.
- Idea is too vague: stay in understanding mode; ask one narrowing question at a time.
- No bounded plan exists for grill/debate: ask for one; do not invent opposing positions.
- If the user asks for task sequencing, state that it is outside this skill's scope.
- If the user asks for generic technology research, route to
researching-web. - If the user asks for docs-only work, route to
documenting-code. - If the user stops mid-session, offer
BRAINSTORM PAUSEDor a short design note.