Skip to content

Security

Style Dataset Lab is a local-only tool. It runs on your machine, talks to a local ComfyUI server, and never makes outbound network requests.

ComponentRiskMitigation
ComfyUI serverListens on 127.0.0.1:8188. If exposed to a network, an attacker could inject workflow payloads.Keep the default localhost binding. Never bind to 0.0.0.0 or a public interface.
Model weights.safetensors, .ckpt, .pth, .bin files loaded by ComfyUI could contain malicious pickle payloads (especially .ckpt and .pth).Only load weights from sources you trust. Prefer .safetensors format, which is pickle-free by design.
Prompt packsJSON files that define generation parameters. A malicious prompt pack could attempt path traversal in output paths.Scripts resolve paths relative to the repo root and do not follow symlinks.
Record filesJSON files with judgment and provenance data. Not executable.No risk beyond data integrity.
  • No telemetry. Zero usage data collection. Zero outbound requests beyond the local ComfyUI API.
  • No authentication. The local ComfyUI API has no auth layer. This is acceptable for localhost but dangerous if the server is network-exposed.
  • No remote storage. All images, records, and exports stay on your local filesystem.
  • No code execution from records. Record JSON is read as data, never evaluated or executed.

ComfyUI loads model weights at startup and during workflow execution. The trust chain is:

  1. Checkpoints (.safetensors, .ckpt) — the base diffusion model. The .safetensors format is safe by design (no arbitrary code). The .ckpt format uses Python pickle and can contain arbitrary code — only load .ckpt files from trusted sources.

  2. LoRAs (.safetensors) — style and subject fine-tunes. Same trust rules as checkpoints.

  3. ControlNet models — structural guidance models. Same trust rules.

  4. IP-Adapter models — style reference models. Same trust rules.

Recommendation: Use .safetensors exclusively. The default generation setup (DreamShaper XL Turbo + ClassipeintXL LoRA) uses only .safetensors files.

DataLocationSensitive?
Generated imagesoutputs/No — synthetic images, no PII
Provenance recordsrecords/No — generation parameters only
Judgment recordsrecords/No — aesthetic scores and explanations
Comparisonscomparisons/No — pairwise preference judgments
Exportsexports/No — training data derived from above
Prompt packsinputs/prompts/No — text prompts and generation config

Large binary files are excluded from version control:

  • outputs/candidates/*.png — raw generations (regenerable)
  • outputs/approved/*.png — curated images (regenerable)
  • outputs/rejected/*.png — rejected images (regenerable)
  • exports/ — training data output (regenerable)
  • *.safetensors, *.ckpt, *.pth, *.bin — model weights

Records (records/*.json) and comparisons (comparisons/*.json) are tracked in git because they represent human judgment that cannot be regenerated.

If you discover a security issue, email 64996768+mcp-tool-shop@users.noreply.github.com. Expected response time: 72 hours. Do not open public issues for security vulnerabilities.