The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted job descriptions (JD) via text input or WebFetch. This creates a surface for indirect prompt injection where malicious instructions embedded in a JD could influence the agent's analysis or output.
Ingestion points: Job description text provided as input or fetched from a URL via WebFetch (SKILL.md).
Boundary markers: Absent. The skill does not define clear delimiters or instructions to ignore embedded commands within the JD text.
Capability inventory: Includes Bash (for curl), Write (for report generation), and WebFetch (for remote data retrieval).
Sanitization: None. The skill extracts raw strings from the JD and uses them in subsequent steps without validation or escaping.
[COMMAND_EXECUTION]: The skill uses the Bash tool to execute a curl command that updates a local API. This command includes variables like {Company} and {Role} which are extracted directly from the untrusted JD. If these fields contain shell metacharacters (e.g., backticks, semicolons), it could potentially lead to command injection depending on how the agent constructs the shell string.

score-jd