brainstorm
Skill9 starsChanged 3 months agoMITView on GitHub
Use when the user has a complex idea that needs design exploration before story decomposition.
189 other projects use a version of this.
Adopted by
Most recently taken up by
- crysix/skillstheir own version, first seen Nov 15, 2026
- fullstackhouse/skillstheir own version, first seen Nov 15, 2026
- chuanglibs/Claude-Code-Game-Studios-Codextheir own version, first seen Nov 15, 2026
- yagizdo/quivertheir own version, first seen Nov 15, 2026
- automagik-si/genietheir own version, first seen Nov 15, 2026
- ZaxbyHub/opencode-swarmtheir own version, first seen Nov 15, 2026
Used with
Projects with it also use playtest-report, team-polish and 10 more.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
CoralGameSkillplaytest-report
Generates a structured playtest report template or analyzes existing playtest notes into a structured format. Use this to standardize playtest feedback collection and analysis.
Also in 9 projects that use this
CoralGameSkillteam-polish
Orchestrate the polish team: coordinates performance-analyst, technical-artist, sound-designer, and qa-tester to optimize, polish, and harden a feature or area for release quality.
Also in 11 projects that use this
lilangMaxSkillsoak-test
Generate a soak test protocol for extended play sessions. Defines what to observe, measure, and log during long play sessions to surface slow leaks, fatigue effects, and edge cases that only appear after sustained play. Primarily used in Polish and Release phases.
Also in 6 projects that use this
lilangMaxSkillteam-live-ops
Orchestrate the live-ops team for post-launch content planning: coordinates live-ops-designer, economy-designer, analytics-engineer, community-manager, writer, and narrative-director to design and plan a season, event, or live content update.
Also in 6 projects that use this
- +8 more from CoralGame
Other versions189 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 189
- Donchitos/Claude-Code-Game-Studios★ 39ksame name
- codeaholicguy/ai-devkit★ 1.7ksame name
- jtydhr88/ComfyTV★ 1.2ksame name
- event-driven-io/emmett★ 562same name
- ZaxbyHub/opencode-swarm★ 531same name
- zaxbysauce/opencode-swarm★ 519same name
- automagik-dev/genie★ 353same name
- automagik-si/genie★ 353same name
- namastexlabs/automagik-genie★ 351same name
- bkywksj/knowledge-base★ 340same name
179 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository56 other files it gives its agents.
- .claude/agents/CLAUDE.mdCLAUDE.md
- .claude/hooks/CLAUDE.mdCLAUDE.md
- CLAUDE.mdCLAUDE.md
- .claude/prompts/CLAUDE.mdCLAUDE.md
- .claude/rules/CLAUDE.mdCLAUDE.md
- .claude/skills/CLAUDE.mdCLAUDE.md
- docs/CLAUDE.mdCLAUDE.md
- vision/CLAUDE.mdCLAUDE.md
- llms.txtllms.txt
- scaffold/llms.txtllms.txt
- architecture-checkSkill
- backlog-reviewSkill
- bootstrapSkill
- buildSkill
- claude-sense-checkSkill
- code-qualitySkill
- commitSkill
- continueSkill
- custom-hooksSkill
- dashboardSkill
- debug-sessionSkill
- deploySkill
- discoverSkill
- doctorSkill
- fix-issueSkill
- framework-upgradeSkill
- handoffSkill
- help-meSkill
- ideateSkill
- lanesSkill
- manual-testSkill
- merge-downSkill
- merge-upSkill
- optimizeSkill
- parallel-workSkill
- performance-checkSkill
- phase-reviewSkill
- pr-statusSkill
- quality-checkSkill
- quickstartSkill
- refine-loopSkill
- researchSkill
- retrospectiveSkill
- security-auditSkill
- skill-createSkill
- skill-evalSkill
- sprint-endSkill
- sprint-startSkill
- story-cycleSkill
- testing-cycleSkill
- test-validatorSkill
- UAT-cycleSkill
- undo-workSkill
- uninstallSkill
- weekly-maintenanceSkill
- wrap-upSkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
16 rules. 0 are common, 13 are distinctive, and 3 are very specific to this project.
0
Common
Most of the 157 edited versions we read keep these.
13
Distinctive
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 43
Persona context: Read docs/context/personas.md if it exists.
- Distinctive: line 45
Ask clarifying questions if the idea is vague.
- Very specific: line 61
Before generating alternatives, scan docs/adr/ YAML frontmatter for status: accepted records with tags relevant to this idea's domain.
- Distinctive: line 70
Before proposing alternatives, research them with web evidence.
- Distinctive: line 120
If personas were loaded: Add a Persona Fit line to each approach's preview, evaluating how well the approach serves each persona:
- Distinctive: line 134
Do any proposed approaches violate ground rules?
- Distinctive: line 138
If the design involves API endpoints, authentication flows, data storage, service-to-service communication, or user input processing, run a …
- Distinctive: line 166
Do NOT invoke /ideate, write any stories, or begin implementation until the user has approved a design approach.
- Very specific: line 171
After user approves the design, save the exploration to docs/brainstorms/<topic-slug>.md with frontmatter:
- Distinctive: line 194
After user approves the design:
- Distinctive: line 198
For unknowns: Create spike stories to resolve open questions first
- Distinctive: line 215
NEVER jump to a solution without exploring alternatives
- Distinctive: line 216
NEVER start implementation during brainstorming
- Distinctive: line 217
ALWAYS present at least 2 approaches with tradeoffs
- Distinctive: line 218
ALWAYS identify risks before recommending an approach
- Distinctive: line 219
Keep the design brief concise — save detail for story-level planning
“Common” is measured: kept by at least 6 in 10 of the 157 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
brainstorm
Skill metrics: Emit a start event to the activity log:
echo "{\"type\":\"skill\",\"event\":\"start\",\"skill\":\"brainstorm\",\"ts\":\"$(date -u +%Y-%m-%dT%H:%M:%SZ)\"}" >> docs/sessions/.activity-log.jsonlBrainstorming: $ARGUMENTS
Plan Mode (Optional)
For complex or high-stakes brainstorms, consider using Plan Mode for Phases 1-4. This adds an extra layer of safety by preventing premature implementation during design exploration.
When to use: The idea is architecturally significant, affects multiple systems, or the user explicitly requests a careful exploration.
How: Enter Plan Mode before Phase 1. Remain in Plan Mode through Phase 4 (risk identification). Exit Plan Mode before Phase 5 (design presentation) so you can interact with the user for approval.
1. Explore the Problem Space
Understand the idea deeply before proposing solutions:
- What problem does this solve? Who benefits?
- What are the constraints (technical, business, time)?
- What similar functionality already exists in the codebase?
- What are the non-negotiable requirements vs nice-to-haves?
Product requirements context: Read docs/reference/PRD_SUMMARY.md if it exists. Use Section 3 (success criteria) as evaluation criteria for approaches in Phase 3. Use Section 7 (scope boundaries) to ensure no approach violates stated non-goals or implementation boundaries. Use Section 6 (NFRs) as constraints — e.g., an approach that can't meet the performance or accessibility thresholds is disqualified.
Persona context: Read docs/context/personas.md if it exists. Note each persona's goals, evaluation criteria, and failure scenarios — these inform approach scoring in Phase 3.
Ask clarifying questions if the idea is vague. Do NOT proceed with assumptions.
2. Research the Codebase
Investigate relevant existing code:
- Existing patterns and conventions that a solution must follow
- If
docs/context/system-patterns.mdexists and is populated, load it. In Phase 3, score each approach's Pattern Fit: does it follow or diverge from established implementation patterns, error handling strategy, and testing conventions? Approaches that align are lower-risk. Approaches introducing new patterns must justify the divergence and note which sections of system-patterns.md would need updating. - Related features that this idea connects to
- Architecture constraints from
docs/architecture/ARCHITECTURE.md - Ground rules from
docs/reference/GROUND_RULES.md(if exists) — MUST rules are hard constraints on any proposed approach - Technology limitations or capabilities
2.5. Check Existing Architecture Decisions
<IF condition="docs/adr/ contains accepted ADR files">
Before generating alternatives, scan `docs/adr/` YAML frontmatter for `status: accepted` records with tags relevant to this idea's domain. Treat accepted ADRs as constraints on the design space:
- **Exclude** any approach that matches a `rejected-options` value in an accepted ADR — note the ADR reference and why it was rejected
- **Respect** accepted decisions as fixed constraints — don't propose alternatives that contradict them
- **Check `Reconsider when`** conditions — only if circumstances have materially changed since the ADR was accepted, note this and present the reopened option alongside its history
- Present any relevant ADRs in Phase 3 as context: "Constrained by ADR-NNNN: [title]"
</IF>2.6. Research Alternatives (Optional — Recommended)
Before proposing alternatives, research them with web evidence. This step makes alternatives evidence-backed rather than relying solely on training data.
Compose the deep-research methodology (.claude/prompts/deep-research.md) at STANDARD depth:
- Query: Generated from the 2-3 potential approaches identified during problem exploration
- Sub-questions (one per approach):
- "Real-world implementations of [approach 1] for [problem domain]"
- "Tradeoffs and pitfalls of [approach 2] in production"
- "Libraries/tools commonly used for [approach 3] in [detected stack]"
- Output format:
decision-input(compact, one per approach)
Research is dispatched in parallel — one subagent per approach.
Integrate findings into Phase 3 alternatives: each approach's pros/cons now include real-world evidence and citations where applicable.
Skip when: User says "no research" or "I know the approaches", or the brainstorm is about internal refactoring with no external technology choices.
3. Propose Alternative Approaches (ADR-Informed)
Present 2-3 distinct approaches using AskUserQuestion with markdown previews for visual comparison.
For each approach, create an option with:
- label: Short approach name (e.g., "Event-Driven", "REST API", "Monolith-First")
- description: One-line tradeoff summary (e.g., "Flexible but adds message broker dependency")
- markdown: Rich preview containing:
## [Approach Name] **How:** [2-3 sentence description] **Architecture:** [ASCII diagram or file structure tree] **Pros:** - [benefit 1] - [benefit 2] **Cons:** - [tradeoff 1] - [tradeoff 2] **Complexity:** [Low/Medium/High] — [N files, N new deps] **PRD Fit:** [How well this approach meets PRD success criteria and NFRs] **Pattern Fit:** [Aligns with / Extends / Diverges from established patterns]
If PRD was loaded: Score each approach against PRD success criteria (Section 3) and NFRs (Section 6). An approach that can't meet a stated NFR (e.g., performance target, accessibility requirement) should be flagged as non-viable.
If system-patterns were loaded: Add a Pattern Fit line to each approach. Score: Aligns (follows all documented patterns), Extends (adds a new pattern consistent with existing ones), or Diverges (contradicts a documented pattern — must justify). Approaches that Diverge carry higher implementation risk and require updating system-patterns.md.
If personas were loaded: Add a Persona Fit line to each approach's preview, evaluating how well the approach serves each persona:
- Apply each persona's EVALUATES BY questions to the approach
- Flag approaches that would trigger a persona's FRUSTRATIONS (e.g., an approach requiring complex configuration when the primary persona is non-technical)
- Note when approaches differ by persona fit: "Option A serves P1 better (speed); Option B serves P2 better (simplicity)"
- The primary persona (★) has tiebreaker weight when approaches trade off between personas
State which approach you recommend and why. The user selects their preferred approach.
4. Identify Risks and Open Questions
- What could go wrong with the recommended approach?
- What unknowns remain? (Suggest spikes for unknowns)
- Are there security, performance, or scalability concerns?
- Does this require an ADR (Architecture Decision Record)?
- Do any proposed approaches violate ground rules? A MUST violation disqualifies an approach; a SHOULD violation is a risk to flag.
STRIDE-Light Threat Analysis (for designs with trust boundaries)
If the design involves API endpoints, authentication flows, data storage, service-to-service communication, or user input processing, run a quick STRIDE pass on each component in the data flow:
| Threat | Question | Example |
| Spoofing | Can an attacker impersonate a legitimate user or service? | Missing auth tokens, weak session management |
| Tampering | Can data be modified in transit or at rest? | Unsigned API payloads, unvalidated webhooks |
| Repudiation | Can an action be denied without evidence? | Missing audit logs for sensitive operations |
| Information Disclosure | Can sensitive data leak? | Verbose errors, PII in logs, secrets in config |
| Denial of Service | Can the system be overwhelmed? | Missing rate limiting, unbounded queries |
| Elevation of Privilege | Can a user gain unauthorized access? | IDOR, missing role checks, privilege escalation |
This is not full threat modeling — it's a 2-minute structured checklist. Include findings in the "Risks" section of the design brief. Any STRIDE finding should become a security acceptance criterion when the design is decomposed into stories by /ideate.
5. Present Design for Approval
Summarize the recommended approach in a design brief:
### Design: [Idea Title]
**Problem:** [One sentence]
**Approach:** [Recommended approach name]
**Key decisions:** [Bullet list]
**Risks:** [Bullet list]
**Estimated scope:** [Small: 1-2 stories / Medium: 3-5 / Large: 6+]<HARD-GATE>
Do NOT invoke /ideate, write any stories, or begin implementation until the user has approved a design approach. The purpose of brainstorming is exploration, not execution.
</HARD-GATE>6. Persist Design Document
After user approves the design, save the exploration to docs/brainstorms/<topic-slug>.md with frontmatter:
---
title: "<Idea Title>"
date: <YYYY-MM-DD>
status: decided # explored | decided | abandoned
decision: "<Chosen approach name>"
---Include the design brief (from Phase 5), the approaches explored (from Phase 3), and the risks identified (from Phase 4). This document is referenced by /ideate when the idea becomes a story.
If the decision is architecturally significant (from Phase 4 assessment): also create an ADR using docs/adr/TEMPLATE.md. The brainstorm document captures the exploration; the ADR captures the decision in machine-parseable format with rejected alternatives and compliance checks. Map the brainstorm's approaches to the ADR's Alternatives Considered section (chosen → ✅, rejected → ❌ with rationale and reconsider-when conditions).
Skill metrics: Emit a completion event:
echo "{\"type\":\"skill\",\"event\":\"end\",\"skill\":\"brainstorm\",\"outcome\":\"success\",\"ts\":\"$(date -u +%Y-%m-%dT%H:%M:%SZ)\"}" >> docs/sessions/.activity-log.jsonl7. Next Steps
After user approves the design:
- For immediate implementation: Invoke
/ideatewith the approved approach to decompose into stories - For complex designs: The design doc in
docs/brainstorms/serves as reference during implementation - For unknowns: Create spike stories to resolve open questions first
Example
Input: /brainstorm "real-time notifications"
Output: 3 approaches explored (WebSockets, SSE, polling)
Recommended: SSE — simpler, sufficient for one-way notifications
Risks: browser support for SSE reconnection, scaling beyond 1000 connections
Next Steps:
→ /ideate "real-time notifications using SSE" — decompose into stories
→ /handoff — if ending the sessionRules
- NEVER jump to a solution without exploring alternatives
- NEVER start implementation during brainstorming
- ALWAYS present at least 2 approaches with tradeoffs
- ALWAYS identify risks before recommending an approach
- Keep the design brief concise — save detail for story-level planning
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

