The Agent Skills Directory

[COMMAND_EXECUTION]: The skill orchestrates several system utilities and local project scripts to analyze and modify the codebase.\n
Utilizes ripgrep (rg) for high-performance pattern matching and jq for manipulating JSON data in scripts/scan.sh and scripts/apply.sh.\n
Uses git across multiple scripts to validate the repository's clean state, parse the current HEAD, and perform file restorations in case of failure.\n
Employs python3 for date calculations to ensure scan data remains fresh before application.\n
Executes project-specific verification commands, such as npm run lint and npx tsc, using eval within scripts/verify.sh.\n- [EXTERNAL_DOWNLOADS]: The skill uses well-known package runners to fetch standard tools from official registries.\n
scripts/apply.sh invokes npx --yes jscodeshift@latest to obtain the migration tool from the npm registry.\n
scripts/verify.sh invokes npx tsc (depending on configuration) to run type-checking.\n- [INDIRECT_PROMPT_INJECTION]: The skill has a data ingestion surface that processes untrusted local source code, which is then summarized in a report for the agent and user.\n
Ingestion points: scripts/scan.sh reads code from .ts, .tsx, .js, and .jsx files in the provided repository root.\n
Boundary markers: The dry-run report generated by scripts/report.sh uses markdown tables and escapes pipe characters to maintain structural integrity.\n
Capability inventory: The skill is capable of modifying files via jscodeshift and executing shell commands via the configured verification scripts.\n
Sanitization: Extracted code snippets are truncated to a maximum of 120 characters to minimize the amount of external content displayed in the report.

nuqs-codemod-runner