B — Some risk signals Graded October 16, 2026Minor concerns worth a quick check.Based on public adoption and transparency signals as of October 16, 2026. Our opinion, not a code audit or security guarantee.
AI agent skill · Developer tools
editor
Open-source 3D architectural editor with a local CLI, MCP tools, and practical workflows for humans and AI agents.
Bottom line: Signal-based grade (4/7 points): very widely starred on GitHub, public source code. This reflects adoption and transparency, not a code review.
- Popularity score
- 49/100
- Rank
- #138 of 500
- GitHub stars
- 24.9k
What it does
A 3D architectural editor built for AI agents and humans, featuring a local CLI, MCP tools, and workflows designed to integrate with AI systems. Enables programmatic 3D model creation and manipulation through both command-line and agent-compatible interfaces.
Who it’s for
Builders and enterprises automating architectural design workflows, and AI agents that need to create or modify 3D models as part of broader automation tasks. Teams seeking local-first, open-source alternatives to cloud-dependent design tools.
Copy-paste setup
Check these steps against the project’s own instructions before running anything.
Follow the setup steps in the project’s official README: https://github.com/pascalorg/editor (opens external site)
Risks
- Local execution risk: CLI execution on user machines could be exploited if agent prompts are not sanitized; 3D model parsing may be vulnerable to malformed file attacks.
- Supply chain: Early-stage project (high GitHub stars but low adoption signals); npm package integrity depends on maintainer account security.
- Data exposure: 3D models and architectural data are saved locally; ensure agent workflows don't exfiltrate sensitive design files.
- Prompt injection: MCP tools accepting free-form agent input may be vulnerable to instruction injection if not properly scoped.
- Permissions scope: Unclear whether CLI runs with elevated privileges; verify principle of least privilege in deployment.
How to stay secure
Least-privilege permissions
- Run the CLI with non-root user accounts; verify no
sudorequirement for normal model editing/viewing operations - Restrict file system access to project directories only; use OS-level ACLs to prevent agent access to
~/.ssh,~/.aws, or other credential stores - Audit MCP tool definitions to ensure they don't accept shell metacharacters; whitelist allowed file paths and reject
..traversal patterns - Disable or sandbox any CLI features that invoke external tools (compilers, renderers) if not strictly necessary for the workflow
Network isolation
- Run the editor and agent in an air-gapped environment or container with no outbound internet access except for explicitly whitelisted package registries during setup
- Block the agent process from accessing localhost ports beyond those required for the editor's local API
- Verify npm dependencies are installed from a private mirror or locked lockfile; use
npm ciinstead ofnpm installin agent deployments - Disable auto-update features; require manual version pinning and security review before upgrades
Monitoring
- Log all CLI invocations triggered by agents, including arguments and file paths accessed; alert on unexpected commands or repeated failures
- Monitor for file system changes outside expected project directories; flag bulk file reads or writes to
/tmp, home directories, or archives - Track MCP tool calls for prompt injection patterns (e.g., embedded
&&, backticks,$(...), or control characters); implement input validation logging - Set up alerts for npm package metadata changes (version bumps, new maintainers, dependency additions) on this early-stage project
Red flags to watch for
- Agent prompts requesting raw CLI access, file globbing, or piping output to external commands
- MCP tools accepting unsanitized user input directly in file names or command strings
- 3D model files from untrusted sources (especially
.obj,.gltf, or custom formats); validate file headers and size before parsing - Requests to export or upload architectural designs to cloud storage, external APIs, or email; confirm such flows are explicitly user-initiated and logged
Alternatives
- Three.js (opens external site) – Mature JavaScript 3D library with broad ecosystem support, better for web-based workflows.
- Blender Python API (opens external site) – Industry-standard, feature-complete, but heavier footprint and steeper learning curve.
- OpenSCAD (opens external site) – Lightweight, code-driven CAD; simpler but less visual than editor.
- Spline (opens external site) – Modern, AI-friendly 3D editor, but cloud-hosted and proprietary.