The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data by scanning and reading existing blog files to determine content freshness and topic cluster status. An attacker could embed malicious instructions within the markdown or HTML files of a blog to influence the agent's behavior when the calendar is generated.
Ingestion points: Step 1 and Step 2.5 of the workflow in SKILL.md instruct the agent to use Glob and Read on *.md, *.mdx, and *.html files.
Boundary markers: There are no explicit instructions or delimiters provided to prevent the agent from obeying instructions found within the scanned files.
Capability inventory: The skill has access to Bash, Write, Read, and WebFetch tools, which could be abused if an injection is successful.
Sanitization: The skill does not define any sanitization or validation routines for the data extracted from the blog posts.
[COMMAND_EXECUTION]: The skill is configured with powerful tools including Bash, Grep, and Glob. These tools are used to search for specific frontmatter fields (like lastUpdated or date) and scan directory structures. While these are aligned with the skill's purpose of blog analysis, they represent a significant capability that could be targeted by command injection if file paths or content were maliciously crafted.

blog-calendar