The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes UI context data from IDA Pro, which could potentially contain untrusted strings from binaries under analysis. This represents an indirect prompt injection surface inherent to the tool's intended use in a reverse-engineering environment. Ingestion points: UI state capture via the get_ui_context_json() function in SKILL.md. Boundary markers: Not present in the captured context handling logic. Capability inventory: The skill utilizes standard system tools including Bash, Read, Glob, and Grep. Sanitization: Data is retrieved and presented as structured JSON without explicit filtering of user-controlled strings.

ui-context