optimize-tsl
Warn
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: MEDIUMDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [DYNAMIC_EXECUTION]: The
scripts/benchmark.mjsutility uses dynamicimport()to load and execute arbitrary JavaScript or TypeScript code from a path provided as a command-line argument. - Evidence: The line
const { default: createFixture } = await import(pathToFileURL(fixturePath).href)inscripts/benchmark.mjsallows the execution of arbitrary fixture logic. - [DYNAMIC_EXECUTION]: The benchmarking script dynamically resolves and imports project dependencies (
webgpu,three) using computed paths based on the user's environment. - Evidence: The
loadfunction inscripts/benchmark.mjsusesrequire.resolveandimport()to pull in external modules at runtime:const load = (name) => import(pathToFileURL(require.resolve(name)).href). - [INDIRECT_PROMPT_INJECTION]: The skill's primary function is to ingest, analyze, and refactor Three.js Shading Language (TSL) code provided by the user, creating a vulnerability surface where malicious instructions could be embedded in the data.
- Ingestion points: The agent reads user-provided TypeScript/TSL source code from the project environment or user prompt.
- Boundary markers: No specific delimiters or instructions to ignore embedded commands are present in the instructions to protect the agent during analysis.
- Capability inventory: The agent has access to the
scripts/benchmark.mjstool, which possesses file-system write privileges and code execution capabilities. - Sanitization: There is no evidence of sanitization or safety validation for the input shader code before it is refactored or benchmarked.
Audit Metadata