The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to fetch metadata (titles, descriptions, summaries) from external, potentially untrusted URLs using the extract-meta.js script. Content retrieved from these sources is interpolated into the blog post frontmatter and body. Maliciously crafted meta tags or page titles could contain hidden instructions targeting the AI agent's logic during the subsequent file editing or command execution phases.
Ingestion points: extract-meta.js parses HTML content from external URLs provided as arguments.
Boundary markers: The skill instructions do not specify explicit delimiters or warnings for the agent to ignore instructions embedded in the fetched metadata.
Capability inventory: The skill has access to Bash, Write, Edit, and Read tools, providing a surface for actions if an injection is successful.
Sanitization: Content is trimmed and dates are normalized, but there is no specific filtering for prompt injection strings.
[COMMAND_EXECUTION]: The skill involves running shell commands, specifically node extract-meta.js <url> and go run ./cmd/hydrate <kind> <slug>. These commands are used to process the blog post content and metadata within the local environment.
[EXTERNAL_DOWNLOADS]: The extract-meta.js script utilizes the http and https modules to perform network requests to external URLs to extract site metadata for linkposts.

blogpost-creator