The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input, including travel dates, destinations, and uploaded itinerary tables, to generate content for the roadbook HTML page.
Ingestion points: User data enters the context through interaction.ask and uploaded files as mentioned in Step 1 of SKILL.md.
Boundary markers: The skill lacks explicit boundary markers or instructions to isolate user-provided data from its planning and execution logic.
Capability inventory: The skill executes Python scripts for HTML generation (build_roadbook.py) and utilizes the lark-cli tool for deploying files to a remote service.
Sanitization: The script scripts/build_roadbook.py implements html.escape() on all dynamic data fields before they are rendered into the HTML template, mitigating potential cross-site scripting (XSS) vulnerabilities.
[EXTERNAL_DOWNLOADS]: The skill references the installation of the qrcode Python package from a public registry to support the generation of QR codes for the published itinerary links.
[COMMAND_EXECUTION]: The workflow requires the execution of several shell commands, including Python scripts for building the roadbook and a deployment tool (lark-cli) to host the generated itinerary on an external platform.

travel-roadbook