The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The SKILL.md file provides installation instructions that use a piped shell command: curl -fsSL https://raw.githubusercontent.com/coollabsio/cloudflare-cli/main/scripts/install.sh | bash. This pattern allows for arbitrary code execution from a remote source that is not on the trusted vendor list, bypassing package manager verification.
[EXTERNAL_DOWNLOADS]: The skill references several external resources for installation and updates, including go install github.com/coollabsio/cloudflare-cli@latest, git clone from a non-trusted repository, and the cf update command which triggers runtime downloads of binary executables.
[COMMAND_EXECUTION]: The skill relies on the execution of a third-party binary (cf) which performs sensitive operations, including writing API tokens to the local filesystem at ~/.cloudflare/config.yaml and modifying system PATH variables via export PATH instructions.
[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface as it ingests untrusted data from the Cloudflare API (DNS records, comments, and search results). These fields could contain malicious instructions designed to influence the agent's subsequent actions.
Ingestion points: cf dns list, cf dns find, and cf dns get commands read record names, content, and comments from the Cloudflare API (SKILL.md).
Boundary markers: None identified; the skill does not instruct the agent to ignore instructions embedded within the DNS data.
Capability inventory: The skill has the capability to write to the filesystem (saving credentials), execute shell commands, and perform network requests (SKILL.md).
Sanitization: No evidence of sanitization or validation of the content retrieved from the DNS records before it is presented to the agent context.

cloudflare-cli