The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection because it reads and processes content from an external source (the Bear Blog dashboard).
Ingestion points: The skill reads data from div#header_content and textarea#body_content when editing posts or listing content in examples/browser-api-reference.md.
Boundary markers: No specific boundary markers or instructions to ignore embedded commands are present in the processing logic.
Capability inventory: The skill utilizes a browser tool capable of navigation, element interaction, and arbitrary JavaScript execution via evaluate.
Sanitization: No sanitization or validation of the retrieved content is described before it is processed by the agent.
[COMMAND_EXECUTION]: The skill uses the browser tool's evaluate function to execute JavaScript within the context of the web page.
Evidence: Examples in examples/browser-api-reference.md demonstrate using POST /act {"kind": "evaluate", "fn": "..."} to manipulate the DOM, override browser dialogs (window.confirm), and extract data.
Context: These executions are localized to the browser environment and are standard for automated web interaction tasks defined in the skill's workflow.

bearblog