The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes regular expression rules and test fixtures from external JSON files provided by the user or sourced from third-party character cards. This creates an attack surface where malicious instructions or highly complex patterns could be embedded in the data to influence the agent's behavior or cause resource exhaustion.
Ingestion points: The scripts validate-tavern-regex.mjs, run-regex-fixtures.mjs, and trace-render-pipeline.mjs read user-supplied JSON files from the filesystem using fs.readFileSync.
Boundary markers: The skill does not provide specific delimiters or safety instructions when processing the content of these files, relying on the agent to treat them as plain data artifacts.
Capability inventory: The skill possesses the capability to read local files, dynamically compile regular expressions via the RegExp constructor, and execute string replacements.
Sanitization: While the utility performs schema and type validation (e.g., checking if fields are non-empty strings), it does not include sanitization logic to detect malicious regex patterns, such as those designed for Regular Expression Denial of Service (ReDoS).
[DYNAMIC_EXECUTION]: The script scripts/validate-tavern-regex.mjs utilizes the RegExp constructor to compile patterns from string input at runtime. While this is standard for a regex validation tool, it represents a dynamic execution surface where external input strings control the behavior of the Node.js regex engine.

sillytavern-render-regex-pipeline