The Agent Skills Directory

[REMOTE_CODE_EXECUTION] (CRITICAL): The skill's installation process uses a highly dangerous pattern where a script is downloaded from the internet and immediately executed with full shell privileges. Evidence: curl -sSL https://canifi.com/skills/firebase/install.sh | bash in SKILL.md. Evidence: curl -sSL https://canifi.com/install.sh | bash in SKILL.md.
[EXTERNAL_DOWNLOADS] (HIGH): The skill depends on files and tools hosted on canifi.com, which is an untrusted third-party domain. There is no verification of the integrity or safety of the content being downloaded.
[CREDENTIALS_UNSAFE] (HIGH): The documentation encourages users to provide sensitive plain-text credentials (SERVICE_PASSWORD) to a third-party tool (canifi-env). Because the tool itself is installed via an unvetted RCE pattern, these credentials are at high risk of exfiltration.
[COMMAND_EXECUTION] (MEDIUM): The skill relies on local command execution via the canifi-env utility and browser automation through Playwright to interact with Google services.
[PROMPT_INJECTION] (LOW): The skill is susceptible to indirect prompt injection because it reads and processes external data from Firebase (Firestore, Auth, Analytics). Ingestion points: Firestore documents, Firebase Auth user records, and Analytics dashboard content. Boundary markers: None identified. Capability inventory: The skill has the ability to execute shell commands and perform web actions via Playwright. Sanitization: No sanitization of the retrieved Firebase data is mentioned.

firebase