adobe-premiere-pro

Pass

Audited by Gen Agent Trust Hub on Oct 23, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes external data formats such as FCPXML, AAF, EDL, and CSV files to automate timeline markers and sequence structures.
  • Ingestion points: Data enters the context via the load_chapters function in gpt.md and through interchange file ingestion described in SKILL.md.
  • Boundary markers: While the Python script uses standard CSV parsing, there are no specific AI-level delimiters to ignore instructions embedded in the metadata of these files.
  • Capability inventory: The skill allows for marker creation, sequence modification, and project metadata updates via Adobe UXP and ExtendScript APIs across SKILL.md and gpt.md.
  • Sanitization: The Python helper script validates numeric input via float() conversion, but text fields (like marker names) are passed directly to the Premiere DOM, which is expected functionality.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 23, 2026, 09:13 PM