The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill implements a template for processing arbitrary text from active editor buffers, which creates an ingestion point for untrusted data.
Ingestion points: The assets/package-template/ExamplePlugin.py script uses self.view.substr(region) to read content from the user's current selection.
Boundary markers: The skill does not implement delimiters or instructions to ignore embedded commands within the processed text.
Capability inventory: The plugin is capable of modifying editor buffers via self.view.replace, and the documentation describes running shell-based packaging commands (python3 -m zipfile) and editor commands for unit testing.
Sanitization: The implementation uses json.dumps() which provides structural encoding for the output string but does not sanitize the semantic content for potential malicious instructions.
[EXTERNAL_DOWNLOADS]: The documentation recommends utilizing external resources for plugin validation.
Evidence: assets/package-template/TEMPLATE.md and references/packaging-and-validation.md suggest downloading the UnitTesting package from the official SublimeText organization on GitHub.

sublime-plugin-development