The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute various wrangler CLI commands for project inspection, resource management, and deployment. It includes specific guidance to use project-local versions via package managers to prevent version mismatch or unintended global state changes.
[EXTERNAL_DOWNLOADS]: The skill references and fetches content from Cloudflare's official developer documentation (developers.cloudflare.com) to retrieve command syntax, configuration schemas, and migration guides. These references target a well-known technology service and are used for legitimate instructional purposes.
[INDIRECT_PROMPT_INJECTION]: The skill processes project-local files and external documentation to formulate CLI commands and configuration changes, creating a surface for potential indirect injection.
Ingestion points: Project configuration files (package.json, wrangler.jsonc, cloudflare.config.ts) and external documentation from developers.cloudflare.com.
Boundary markers: None specified in the instructions for delimiting external content.
Capability inventory: Shell command execution via wrangler, file system read/write for configuration, and management of remote Cloudflare resources.
Sanitization: The instructions do not define specific sanitization or escaping protocols for data retrieved from external documentation or project files.
[CREDENTIALS_UNSAFE]: The skill implements protective measures for secret handling, explicitly directing the agent to avoid asking for API tokens in chat and to keep sensitive values out of command arguments and logs. It mandates the use of interactive inputs or protected file mechanisms for secret management.

wrangler