clawhub
Warn
Audited by Gen Agent Trust Hub on Feb 18, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS] (MEDIUM): The skill metadata specifies the installation of the
clawhubpackage via npm and interacts withhttps://clawhub.comto fetch resources. - [REMOTE_CODE_EXECUTION] (MEDIUM): The primary function of this skill is to use
clawhub installandclawhub updateto download and execute code from a third-party registry. This introduces a significant risk because the downloaded skills are not audited and can execute arbitrary logic in the agent's environment. - [COMMAND_EXECUTION] (LOW): The skill executes multiple shell commands to interact with the
clawhubCLI for searching, listing, and updating skills. - [DATA_EXFILTRATION] (LOW): The
publishcommand transmits local folders to a remote server. This functionality could be used to accidentally exfiltrate sensitive files if the target directory is not carefully managed by the user. - [INDIRECT_PROMPT_INJECTION] (LOW): Mandatory Evidence Chain: 1. Ingestion points:
clawhub install(SKILL.md). 2. Boundary markers: Absent. 3. Capability inventory: Installation of new executable skills. 4. Sanitization: Absent. This creates a surface where a malicious skill hosted on the registry could inject instructions into the agent's context.
Audit Metadata