The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/analyze-project.js uses child_process.spawn to execute npm, yarn, or pnpm with the --version flag. This functionality is used for environment detection and is restricted to a specific set of hardcoded package manager binaries.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external project sources, such as directory structures, filenames, and package.json content. This data is processed and included in generated reports (Markdown, JSON, Scorecards) without explicit sanitization or boundary markers. Malicious content within a scanned project could potentially be used to influence the behavior of an agent reading the resulting analysis.
Ingestion points: Files and directory names within the user-specified project path are read by scripts/analyze-project.js and the detector modules.
Boundary markers: The scripts/report-generator.js module does not implement explicit delimiters or warnings to separate untrusted project data from the analysis instructions.
Capability inventory: The skill possesses capabilities for recursive file system reading (fs.readdir, fs.readFile) and limited subprocess execution via child_process.spawn.
Sanitization: There is no evidence of content escaping or validation for external data before it is interpolated into the final reports.

architecture-analysis