The Agent Skills Directory

[Indirect Prompt Injection] (LOW): The skill fetches content from external URLs to generate summaries and archive articles. This processes untrusted data which could contain malicious instructions designed to influence the agent's behavior.
Ingestion points: Content fetched from external URLs (Workflow Step 3).
Boundary markers: Absent. There are no instructions to the agent to ignore or delimit instructions found within the fetched resources.
Capability inventory: File system write access to ~/knowledge/, file system read access, and execution of local shell scripts.
Sanitization: Absent. The skill does not describe any validation or escaping of the fetched content before processing.
[Unverifiable Dependencies] (LOW): The skill relies on an external command get_tabs.sh. This script is not provided within the skill's source and its behavior cannot be audited, representing a dependency on an external, unverified component.
[Data Exposure] (LOW): The skill is configured to read from and write to the user's home directory (~/knowledge/). While this is central to its stated purpose, it establishes a local file system footprint that is manipulated based on inputs from external web sources.

knowledge-base