The Agent Skills Directory

[COMMAND_EXECUTION]: The skill performs shell command execution to verify image URLs.
Evidence: In Phase 2, the skill uses curl -sI "<url>" | head -1 to check for HTTP 200 responses.
Risk: The <url> variable is populated from external search results (Pixabay, Unsplash, Pexels). If a search result provides a maliciously crafted URL containing shell metacharacters, it could lead to arbitrary command execution on the host system.
[EXTERNAL_DOWNLOADS]: The skill interacts with external services to retrieve media assets.
Evidence: The workflow fetches image data from well-known services including Pixabay, Unsplash, and Pexels. These are recognized as legitimate sources for the skill's primary purpose.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection due to its processing of untrusted external data.
Ingestion points: Phase 2 retrieves data from web searches and external image hosting platforms.
Boundary markers: No delimiters or specific instructions are provided to the agent to ignore instructions embedded in the external content.
Capability inventory: The skill can execute shell commands (curl), spawn sub-agents (blog-researcher), and write files to the local system.
Sanitization: There is no evidence of sanitization or validation of the data retrieved from external sources before it is used in shell commands or interpolated into the article generation prompt.

blog-write