The Agent Skills Directory

Indirect Prompt Injection: The skill processes information from various external regulatory feeds and manual user inputs. This creates a potential surface for indirect prompt injection, where instructions could be embedded in the regulatory text to influence the agent's summarization or classification tasks.
Ingestion points: External RSS feeds, Federal Register API responses, and user-pasted text as described in the workflow.
Boundary markers: The instructions do not specify the use of delimiters or explicit boundary markers when interpolating external content into the model's context.
Capability inventory: The skill is authorized to write files to the local filesystem (e.g., digests and comment trackers) and perform network read operations via APIs and RSS.
Sanitization: There are no defined sanitization or validation steps for external content before it is processed for classification and summarization.
Local File Access and Modification: The skill is configured to read from and write to specific paths in the user's home directory, specifically within ~/.claude/plugins/config/. While this is functional for maintaining state and configuration, it involves permissions to modify local file content and read watchlist settings.
External Data Retrieval: The skill performs network requests to various external domains listed in the source catalog. While the catalog focuses on reputable government and legal organizations, the mechanism allows the agent to fetch and process data from external sources specified in the user's configuration.

reg-feed-watcher