Back to skill
v0.1.0

Brainstorming

ReviewClawScan verdict for this skill. Analyzed May 1, 2026, 8:01 AM.

Analysis

This is mostly a coherent planning skill, but review it carefully because its visual companion runs an unauthenticated browser/WebSocket server that can be exposed beyond localhost and can influence the agent’s recorded choices.

GuidanceUse this skill if you want a strict design-first workflow. Before installing, be comfortable with local file writes/commits and helper script execution. Keep the visual companion on localhost unless you can secure the port, add `.superpowers/` to `.gitignore`, and clean up old brainstorming session files when finished.

Findings (6)

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

Abnormal behavior control

Checks for instructions or behavior that redirect the agent, misuse tools, execute unexpected code, cascade across systems, exploit user trust, or continue outside the intended task.

Agent Goal Hijack
SeverityLowConfidenceHighStatusNote
SKILL.md
You MUST use this before any creative work ... Do NOT invoke any implementation skill, write any code, scaffold any project, or take any implementation action until you have presented a design and the user has approved it.

The skill deliberately changes the agent’s workflow by requiring a design gate before implementation. This matches the stated brainstorming purpose, but it is broad and can override a user’s desire for quick direct changes.

User impactThe agent may refuse or postpone implementation tasks until it has completed a design/spec process.
RecommendationInstall this only if you want a mandatory design-first workflow; otherwise invoke it manually only when planning is desired.
Tool Misuse and Exploitation
SeverityLowConfidenceHighStatusNote
SKILL.md
Write design doc — save to `docs/superpowers/specs/YYYY-MM-DD-<topic>-design.md` and commit

The skill instructs the agent to mutate the project workspace and create a git commit. That is purpose-aligned for a spec workflow, but it affects repository state.

User impactThe skill can add files and commits to your project during the brainstorming process.
RecommendationAsk the agent to confirm before writing or committing specs if you want tighter control over repository changes.
Agentic Supply Chain Vulnerabilities
SeverityLowConfidenceHighStatusNote
metadata
Install specifications: No install spec — this is an instruction-only skill. Code file presence: 4 code file(s): scripts/helper.js, scripts/server.cjs, scripts/start-server.sh, scripts/stop-server.sh

The registry presentation under-declares the runnable helper scripts included with the skill. The helper code is visible in the artifacts and appears related to the visual companion, so this is a review note rather than evidence of hidden behavior.

User impactA user may expect a purely instructional skill, but using the visual companion requires local script execution.
RecommendationReview the included scripts and ensure Bash/Node are acceptable in your environment before using the visual companion.
Rogue Agents
SeverityLowConfidenceHighStatusNote
scripts/start-server.sh
nohup env BRAINSTORM_DIR="$SCREEN_DIR" ... node server.cjs > "$LOG_FILE" 2>&1 & ... disown "$SERVER_PID"

The visual companion can run as a detached background process. The artifacts disclose this and include shutdown/idle behavior, so it is bounded but still worth noticing.

User impactA helper server may keep running after the initial command returns.
RecommendationUse the provided stop script or foreground mode when you want explicit process control.
Sensitive data protection

Checks for exposed credentials, poisoned memory or context, unclear communication boundaries, or sensitive data that could leave the user's control.

Insecure Inter-Agent Communication
SeverityMediumConfidenceHighStatusConcern
scripts/server.cjs
const HOST = process.env.BRAINSTORM_HOST || '127.0.0.1'; ... function handleUpgrade(req, socket) { const key = req.headers['sec-websocket-key']; if (!key) { socket.destroy(); return; } ... clients.add(socket); } ... fs.appendFileSync(eventsFile, JSON.stringify(event) + '\n');

The companion server bind host is configurable and WebSocket clients are accepted with only the standard handshake key; shown code does not authenticate the browser/client before recording events as user choices. The companion guide also documents using non-loopback binding for remote/containerized setups.

User impactIf the server is reachable by other local or network clients, they could potentially view the brainstorming screen or inject click/choice events that the agent later treats as browser feedback.
RecommendationKeep the server bound to localhost whenever possible; if remote access is needed, add a firewall, tunnel, token, or origin check before trusting browser events.
Memory and Context Poisoning
SeverityLowConfidenceHighStatusNote
visual-companion.md
Selections are recorded to a `.events` file that you read on your next turn. ... mockups persist in `.superpowers/brainstorm/` and survive server restarts.

The visual companion stores user interaction data and generated mockups for later agent use. This is central to the feature, but it creates local persistent context that should not be over-trusted or accidentally committed.

User impactBrowser choices and generated design screens may remain on disk and influence later turns.
RecommendationKeep `.superpowers/` out of version control, delete old brainstorming sessions when done, and treat `.events` as advisory rather than authoritative.