The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes standard command-line tools including aws, terraform, and docker to manage infrastructure and validate configuration files. These actions are within the expected scope of a deployment and configuration tool.
[EXTERNAL_DOWNLOADS]: Interacts with the Cloudflare API (api.cloudflare.com) via curl to perform cache invalidation. This is a well-known service and the operation is central to the skill's stated purpose of managing CDN caches.
[INDIRECT_PROMPT_INJECTION]: The skill analyzes local project configuration files (e.g., vercel.json, wrangler.toml, nginx.conf) to determine deployment settings. While processing external files creates an attack surface for indirect prompt injection, the skill implements a self-healing phase to validate the syntax of generated configurations.
Ingestion points: Scans project files such as wrangler.toml and vercel.json in Phase 1.
Boundary markers: None identified.
Capability inventory: The skill can write configuration files and execute deployment commands (aws, terraform, curl).
Sanitization: Performs syntactic validation and syntax checking (Phase: SELF-HEALING) but does not explicitly describe semantic sanitization of ingested content.

cdn