The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides utility scripts (inspect-project.ts, patch-settings.ts, validate-hooks.ts) for managing Claude Code hook configurations. These scripts are intended to be executed via Bun to read and modify local .claude/settings.json files and verify the presence of hook scripts.
[EXTERNAL_DOWNLOADS]: The check-official-links.ts script performs network operations to code.claude.com and github.com to verify reference documentation links. These operations use the HTTP HEAD method to check URL status without downloading or executing content bodies.
[INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect injection by ingesting data from the local environment, such as package.json, .claude/settings.json, and directory listings. This information is processed by scripts and returned to the agent context.
Ingestion points: scripts/inspect-project.ts reads repository configuration files and file system metadata.
Boundary markers: No specific delimiters are used in script outputs beyond standard JSON formatting.
Capability inventory: The skill is capable of reading local files and writing to the .claude/settings.json configuration file.
Sanitization: Ingested content is handled using standard JSON parsing (JSON.parse).

rekooh-cc