The Agent Skills Directory

[SAFE]: The skill provides instructions for using a custom SQL function get_ui_context_json() to retrieve UI metadata from IDA Pro, such as widget types, window titles, and code addresses. This functionality is consistent with the skill's stated purpose of providing UI context to the agent.
[INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of data from the IDA Pro UI, which may include user-controlled content like function names, comments, or selected text. This constitutes an indirect prompt injection surface.
Ingestion points: Data is ingested via the get_ui_context_json() function described in SKILL.md.
Boundary markers: The instructions provide a Response Template to structure the agent's output, but do not specify explicit delimiters or isolation for the raw UI text within the input.
Capability inventory: The agent is granted access to standard tools including Bash, Read, Glob, and Grep as defined in the YAML frontmatter.
Sanitization: There are no instructions for sanitizing or escaping the text captured from the UI before it is processed by the agent.

ui-context