The Agent Skills Directory

[SAFE]: No security issues were detected in the skill's instructions, scripts, or templates.
[COMMAND_EXECUTION]: The skill uses the xdg-open (or open on macOS) command to display the generated HTML artifacts in the user's browser. This is an expected behavior for the skill's primary function.
[COMMAND_EXECUTION]: The scripts/validate.sh script is provided for self-validation of the skill's integrity. It executes a Python auditing script (audit.py) if found in predefined directories.
[PROMPT_INJECTION]: The skill processes external data such as code, logs, and communications to populate HTML templates, creating an indirect injection surface.
Ingestion points: Operator input, git logs, and workspace files used to populate templates (as described in SKILL.md and references/*.md).
Boundary markers: Absent. The generated HTML artifacts do not utilize specific delimiters to isolate untrusted user data from the template structure.
Capability inventory: The skill can use Read, Write, Edit, Bash, and AskUserQuestion tools.
Sanitization: Basic HTML escaping is implemented in some interactive templates (e.g., templates/20-editor-prompt-tuner.html) to ensure safe rendering of data during live previews.

html-artifact