webgpu_error_capture
Capture WebGPU validation/out-of-memory/internal errors the target app swallows (via device uncapturederror), plus the current device.lost state. Optionally wraps createBuffer/createTexture in error scopes to attribute failures to specific calls. Surfaces the real diagnostics behind
This record as markdown: /tools/io-github-vmoranv-jshookmcp/webgpu-error-capture.md
What webgpu_error_capture does on Jshookmcp
AI agents call webgpu_error_capture to retrieve information from Jshookmcp without modifying anything. It is typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
Why webgpu_error_capture is rated Low
This is a diagnostics/monitoring tool that reads error states and device information from a running WebGPU context. It does not create, modify, delete, or execute code/operations; it merely observes and reports on existing errors and device.lost state. Even the error scope wrapping is a diagnostic wrapper, not an execution of GPU operations.
From the tool's definition The tool 'captures' and 'surfaces' WebGPU errors and device state without modifying or executing operations—it observes error diagnostics that the target app already suppresses.
Attacks that exploit this kind of access
The rule that runs webgpu_error_capture safely
PolicyLayer is an MCP gateway: it sits between your AI agents and Jshookmcp, and checks every tool call against a rule you set before the call runs. Nothing changes on the server itself. For webgpu_error_capture, this is the rule to start with:
webgpu_error_capture is read-only, so it stays allowed. Everything else on the server is denied unless you say otherwise.
The button opens the PolicyLayer dashboard: create your workspace, connect Jshookmcp, apply this rule, and every webgpu_error_capture call is checked against it from then on.
Questions about webgpu_error_capture
Capture WebGPU validation/out-of-memory/internal errors the target app swallows (via device uncapturederror), plus the current device.lost state. Optionally wraps createBuffer/createTexture in error scopes to attribute failures to specific calls. Surfaces the real diagnostics behind. It is categorised as a Read tool in the Jshookmcp MCP Server, which means it retrieves data without modifying state.
Register the Jshook MCP server in PolicyLayer and add a rule for webgpu_error_capture: 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 Jshookmcp. Nothing to install.
webgpu_error_capture is a Read tool with low risk. Read-only tools are generally safe to allow by default.
Yes. Add a rate_limit block to the webgpu_error_capture 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 webgpu_error_capture. 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.
webgpu_error_capture is provided by the Jshook MCP server (@jshookmcp/jshook). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
More on Jshook, and thousands of servers like it.
This server
Across the catalogue