lazyweb-add-inspo-source
Pass
Audited by Gen Agent Trust Hub on Sep 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [DYNAMIC_EXECUTION]: The skill uses a shell script to search for the 'browse' executable across multiple paths, including hidden directories in the user's home folder and the current project. The executable path is stored in a variable and then invoked, which constitutes dynamic execution from a computed path.
- [COMMAND_EXECUTION]: The skill executes a local binary ('browse') to perform browser automation tasks. While this is the intended mechanism for connecting design libraries, executing binaries located in user-writable paths is a significant capability that requires the surrounding environment to be secure.
- [INDIRECT_PROMPT_INJECTION]: The skill navigates to external URLs and takes snapshots of the UI to identify form fields and search results. This process ingests untrusted data from third-party websites into the agent's context, which could potentially be used to influence the agent's behavior via hidden or malicious instructions in the web content.
- Ingestion points: External websites visited during the connection process (SKILL.md); the '~/.lazyweb/libraries.json' configuration file.
- Boundary markers: The skill does not implement specific delimiters or 'ignore' instructions when processing content from website snapshots.
- Capability inventory: Execution of a browser automation tool, network access via that tool, and local file system read/write operations.
- Sanitization: There is no validation or sanitization shown for the library URLs or search queries before they are passed to the browser tool.
Audit Metadata