The Agent Skills Directory

[DYNAMIC_EXECUTION]: The transform.js script (and its .cjs equivalent) uses the new Function constructor to evaluate JavaScript expressions provided via the --map and --filter command-line arguments. This allows for dynamic code generation and execution based on inputs constructed at runtime.
[REMOTE_CODE_EXECUTION]: The safeEval function in transform.js attempts to restrict the evaluation context by shadowing standard globals like Math and JSON. However, this is an incomplete sandbox that does not prevent access to the full Node.js environment (e.g., via process or constructor chains), allowing arbitrary code execution within the process.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external files (JSON, CSV, YAML, etc.) and apply transformations. It lacks input validation or sanitization, which could allow maliciously crafted data to exploit the dynamic execution capabilities of the transformation tool.
Ingestion points: Data is read from files in convert.js, query.js, transform.js, and diff.js using fs.readFileSync based on the provided input path.
Boundary markers: None identified; the instructions do not include delimiters or warnings to ignore instructions embedded within the data.
Capability inventory: The skill possesses file system read/write capabilities (fs) and dynamic code execution (new Function).
Sanitization: No validation or filtering is performed on external content before it is processed or used in evaluations.

data-transform