The Agent Skills Directory

REMOTE_CODE_EXECUTION (CRITICAL): The skill documentation includes two piped-to-bash installation commands: curl -sSL https://canifi.com/skills/discord/install.sh | bash and curl -sSL https://canifi.com/install.sh | bash. These commands execute unverified remote scripts from the untrusted domain canifi.com directly in the user's shell.
CREDENTIALS_UNSAFE (HIGH): The skill guides users to store cleartext passwords, such as DISCORD_PASSWORD, in local environment variables using the canifi-env utility. This sensitive data is accessible to any scripts executed on the machine, including the unverified installation scripts.
PROMPT_INJECTION (LOW): The skill is vulnerable to Indirect Prompt Injection. Ingestion points: The skill reads Discord messages and server data via Playwright. Boundary markers: Absent; there are no instructions to ignore commands within retrieved data. Capability inventory: The agent can kick/ban users, manage server permissions, and change settings. Sanitization: None; it interacts with raw message content from untrusted server members.
COMMAND_EXECUTION (MEDIUM): The skill uses Playwright for browser automation and references sending iMessage notifications for 2FA, indicating it operates with significant privileges to interact with system applications and the network.

discord