Execute arbitrary JavaScript with access to scene, renderer, camera. Returns the result. Runtime only (lost on reload). If the code mutates scene state, ASK the user first: runtime preview or persistent code change? SHADER PATCHING RECIPE — when modifying materials via onBeforeCompile: 1. Reset p...
Risk signalsAccepts freeform code/query input (code)
Part of the Threejs Devtools server.
Free to start. No card required.
AI agents invoke run_js to trigger processes or run actions in Threejs Devtools. Execute operations can have side effects beyond the immediate call -- triggering builds, sending notifications, or starting workflows. Rate limits and argument validation are essential to prevent runaway execution.
run_js can trigger processes with real-world consequences. An uncontrolled agent might start dozens of builds, send mass notifications, or kick off expensive compute jobs. PolicyLayer enforces rate limits and validates arguments to keep execution within safe bounds.
Execute tools trigger processes. Rate-limit and validate arguments to prevent unintended side effects.
{
"version": "1",
"default": "deny",
"tools": {
"run_js": {
"limits": [
{
"counter": "run_js_rate",
"window": "minute",
"max": 10,
"scope": "grant"
}
]
}
}
}See the full Threejs Devtools policy for all 60 tools.
These attack patterns abuse exactly the kind of access run_js gives an agent. Each links to the full case and the policy that stops it:
Other execute tools across the catalogue. The same approach applies to each: rate-limit and validate the arguments.
Execute arbitrary JavaScript with access to scene, renderer, camera. Returns the result. Runtime only (lost on reload). If the code mutates scene state, ASK the user first: runtime preview or persistent code change? SHADER PATCHING RECIPE — when modifying materials via onBeforeCompile: 1. Reset patch flags: mat._curvedWorldPatched = false; mat._windPatched = false; 2. Delete old shader refs: delete mat.userData.shakeShader; stop old animations via mesh.userData._stopShake?.(); 3. ONE onBeforeCompile pass (never chain) — include curved world + your effect together 4. Replace '#include <project_vertex>' with full transform: USE_BATCHING/USE_INSTANCING guards → effect in world space → curved world (worldPos.y -= dist²*0.005) → viewMatrix*worldPos → gl_Position 5. Declare uniforms before main(): shader.vertexShader.replace('void main() {', 'uniform float uTime;\nvoid main() {') 6. Set mat.customProgramCacheKey = () => 'effect-' + Date.now(); — forces recompilation 7. mat.needsUpdate = true; 8. Animate uniforms via requestAnimationFrame loop 9. Use transformed.y for height-based effects (local vertex Y), worldPos for spatial phase Debug only — page reload will reset. No code changes needed.. It is categorised as a Execute tool in the Threejs Devtools MCP Server, which means it can trigger actions or run processes. Use rate limits and argument validation.
Register the Threejs Devtools MCP server in PolicyLayer and add a rule for run_js: allow, deny, rate-limit, or require approval. Point your MCP client at the PolicyLayer proxy URL and the rule is enforced on every call, before it reaches Threejs Devtools. Nothing to install.
run_js is a Execute tool with high risk. Execute tools should be rate-limited and have argument validation enabled.
Yes. Add a rate_limit block to the run_js rule in your PolicyLayer policy. For example, setting max: 10 and window: 60 limits the tool to 10 calls per minute. Rate limits are tracked per agent session and reset automatically.
Set action: deny in the PolicyLayer policy for run_js. The AI agent will receive a policy violation error and cannot call the tool. You can also include a reason field to explain why the tool is blocked.
run_js is provided by the Threejs Devtools MCP server (threejs-devtools-mcp). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Deterministic rules across all 60 Threejs Devtools tools. Per-identity grants. Full audit log. Live in minutes. Nothing to install.
Free to start. No card required.
4,600+ MCP servers and 31,000+ tools scanned and risk-classified.