The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill provides installation instructions that use curl to download pre-compiled binaries from the author's official GitHub releases page (https://github.com/JulienTant/blogwatcher-cli/releases). This is a standard installation procedure for CLI tools and aligns with the skill's documented purpose.
[COMMAND_EXECUTION]: The skill documents numerous commands for the blogwatcher-cli tool. These commands are executed by the agent to manage blog feeds, scan for updates, and mark articles as read/unread. The commands interact with a local SQLite database and external RSS/Atom feeds.
[INDIRECT_PROMPT_INJECTION]: The skill has the surface area for indirect prompt injection because it ingests data from external RSS/Atom feeds and HTML content via scraping. While the skill does not explicitly process this data into LLM instructions, an attacker could potentially embed malicious prompts in a feed that the agent might interpret if it processes the article summaries or content directly. However, standard agent guardrails typically handle such untrusted content.
Ingestion points: RSS/Atom feeds and HTML scraping targets defined by the user.
Boundary markers: Not explicitly present in the provided skill instructions.
Capability inventory: Executes the blogwatcher-cli tool which performs network requests (feed scanning) and file writes (local SQLite database updates).
Sanitization: Relies on the underlying blogwatcher-cli tool's handling of XML/HTML content.
[NO_CODE]: The skill consists primarily of documentation and instructions for using an existing CLI tool; it does not contain its own executable scripts or complex logic.

blogwatcher