The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses a bash wrapper to execute ExtendScript logic via the macOS osascript utility. This is a standard architectural pattern for automating Adobe applications from external processes.
[DYNAMIC_EXECUTION]: The skill's primary function is to generate and execute custom JSX scripts (ae-action.jsx). This allows for flexible automation tailored to specific user requests, following established After Effects scripting guidelines.
[DATA_EXPOSURE]: Local files in /tmp/ and the user's home directory are used for logging and inter-process communication. These files contain operational data and results, not sensitive system credentials or private user data.
[INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection as it processes external files (SRT, CSV) and has the capability to modify the After Effects project. 1. Ingestion points: In srt-import.jsx, comp-from-csv.jsx, and text-export-import.jsx, the skill reads and parses content from user-provided file paths. 2. Boundary markers: The skill relies on the agent's logic to handle file paths; no explicit delimiters or 'ignore' instructions are used within the data files themselves. 3. Capability inventory: Scripts can create/delete layers, import footage, modify properties, and manipulate the Render Queue. 4. Sanitization: srt-import.jsx performs basic HTML tag stripping from subtitle text. While these ingestion points exist, they are inherent to the skill's utility and do not represent a malicious intent.

after-effects