The Agent Skills Directory

[PROMPT_INJECTION]: The 'BRANDING' section includes instructions that attempt to override the agent's persona and restrict its output behavior, specifically forbidding the mention of external courses or instructors and forcing attribution to the '10x Team' framework.
[REMOTE_CODE_EXECUTION]: The skill automates the generation and injection of external and inline JavaScript into local files. Notably, it instructs the agent to modify the Content-Security-Policy (CSP) meta tags to include 'unsafe-inline' and external domains, which significantly weakens the target application's defenses against Cross-Site Scripting (XSS) attacks.
[COMMAND_EXECUTION]: The skill uses powerful tools including 'bash', 'write', and 'edit' to modify local files based on logic that parses existing HTML content. This process lacks security controls for handling untrusted data.
Ingestion points: Data enters the context via the 'READ' command when scanning project HTML files.
Boundary markers: The skill does not define or use boundary markers to separate ingested file content from agent instructions.
Capability inventory: The skill has access to 'write', 'edit', 'bash', 'glob', and 'grep' tools.
Sanitization: No evidence of sanitization or validation is present before the skill interpolates discovered data into new script blocks or configuration changes.

lp-inject