The Agent Skills Directory

[DATA_EXFILTRATION]: The skill reads the Cloudflare API token from a local file (~/.cloudflare_token) or an environment variable (CLOUDFLARE_API_TOKEN) to authenticate requests. These requests are sent exclusively to the official Cloudflare API endpoint (https://api.cloudflare.com). This behavior is consistent with the skill's primary purpose and targets a well-known service.
[COMMAND_EXECUTION]: The skill utilizes several shell scripts that wrap curl and jq commands. The command arguments are generally handled through standard shell variables, and JSON payloads are constructed using jq --arg in scripts like scripts/dns/create.sh and scripts/tunnels/configure.sh, which prevents potential injection into the JSON structure.
[SAFE]: The setup instructions in SKILL.md recommend setting restricted file permissions (chmod 600) on the token file, which is a security best practice for credential management.
[DATA_EXPOSURE]: The skill surface includes processing data returned from the Cloudflare API (e.g., DNS record names, zone details). While this constitutes an attack surface for indirect prompt injection if the API data is malicious, the skill functions as a standard administrative tool and does not interpolate this data into secondary LLM prompts within its own scripts.

cloudflare-2