after-effects

Warn

Audited by Gen Agent Trust Hub on Oct 27, 2026

Risk Level: MEDIUMPERSISTENCEDATA_EXFILTRATIONDYNAMIC_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONPRIVILEGE_ESCALATION
Full Analysis
  • [PERSISTENCE]: The skill provides a specific example for scheduling a nightly rendering pipeline using a crontab entry (0 1 * * * /opt/render/nightly-render.sh) to maintain automated workflows across sessions.\n- [DATA_EXFILTRATION]: Includes a shell script example that uses curl to transmit rendering logs and status summaries to an external Slack webhook URL, which is a common but sensitive network operation.\n- [DYNAMIC_EXECUTION]: Implements dynamic script evaluation via the Adobe Common Extensibility Platform (CEP) CSInterface.evalScript bridge, which allows the execution of ExtendScript strings within the application context.\n- [COMMAND_EXECUTION]: Utilizes system-level commands to control Adobe After Effects, including the aerender CLI for headless rendering, osascript for macOS application control, and direct binary execution with the -r script flag.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external CSV and JSON files to populate motion graphics templates, creating a surface where untrusted data is processed.
  • Ingestion points: readCSV (SKILL.md Step 3) and footage.sourceData (SKILL.md Step 4).
  • Boundary markers: Absent.
  • Capability inventory: aerender CLI execution, file-system access via ExtendScript File objects, and network requests via curl.
  • Sanitization: Absent.\n- [PRIVILEGE_ESCALATION]: Recommends system configuration changes, such as modifying macOS defaults or Windows Registry keys (PlayerDebugMode), to enable loading of unsigned CEP panels for development.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Oct 27, 2026, 12:52 PM