rspeedy-bundle-size
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill includes scripts that ingest and process external data, specifically application source code and build-related JSON files such as stats.json and rsdoctor-data.json. * Ingestion points: Scripts like scan-levers.mjs and mt-leak-analyzer.mjs read file content from the local file system during analysis. * Boundary markers: The processing scripts do not explicitly use delimiters or warnings to ignore potentially malicious embedded instructions in the source code being analyzed. * Capability inventory: The scripts perform file reading, regular expression matching, and console logging. The strip-mt-telemetry-loader.cjs script uses Babel for AST-based code transformation. * Sanitization: No explicit sanitization or filtering of the input code or metadata is implemented before the analysis results are presented.
- [DYNAMIC_EXECUTION]: The skill implements build-time code transformation to achieve optimization goals. * The strip-mt-telemetry-loader.cjs script utilizes @babel/core to dynamically transform JavaScript source code during the build process, replacing specific telemetry method calls with void expressions. * The gate.awk script provides an AWK-based method to modify source code by injecting thread-specific guards into function bodies.
- [EXTERNAL_DOWNLOADS]: The skill references and links to several external source code repositories and documentation sites. * References official GitHub repositories for the Lynx ecosystem, including the lynx-stack, Rsbuild, Rspack, and Rsdoctor projects. These resources are provided for technical reference and are associated with the well-known organizations developing the underlying toolchain.
Audit Metadata