The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill installs the blogwatcher CLI tool from a third-party GitHub repository (github.com/Hyaxia/blogwatcher) which is not categorized as a trusted source.
[COMMAND_EXECUTION]: The skill invokes the blogwatcher CLI tool with various subcommands to manage and scan feeds, involving shell command execution.
[PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection via the external RSS/Atom feeds it processes.
Ingestion points: The blogwatcher scan and blogwatcher articles commands retrieve content from external, untrusted RSS/Atom feed URLs.
Boundary markers: The SKILL.md file contains explicit defensive guardrails that instruct the agent to restrict command execution, validate URLs, and prevent data exfiltration.
Capability inventory: The agent can execute shell commands through the blogwatcher tool and interact with the local filesystem within the working directory.
Sanitization: No explicit sanitization or filtering of the feed content (such as HTML stripping or instruction filtering) is performed before the data is processed by the agent.

blogwatcher-hardened