The Agent Skills Directory

Prompt Injection (LOW): The skill is susceptible to indirect prompt injection because it ingests and acts upon data from an external source without explicit sanitization.
Ingestion points: The output from the /check-landing primitive in Step 1.
Boundary markers: Absent; there are no delimiters used to separate the ingested audit findings from the skill's core logic.
Capability inventory: The skill possesses the ability to write to the file system (app/page.tsx) and execute shell commands (git, grep, lighthouse).
Sanitization: None; the skill does not validate or escape the content of the audit report before using it to drive the code generation process.
Command Execution (SAFE): The skill utilizes standard developer utilities such as git for branching and grep for string verification. These commands are used for internal workflow logic and do not process unvalidated user input in a way that allows for arbitrary command injection.
File System Access (SAFE): The skill modifies specific application files (app/page.tsx) to implement UI fixes. This behavior is transparent, documented, and restricted to the intended scope of the marketing landing page.

fix-landing