subtitle-refine

Pass

Audited by Gen Agent Trust Hub on Apr 12, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: The skill instructions are focused on task-specific constraints for subtitle refinement and do not contain patterns designed to bypass agent safety filters.
  • [DATA_EXFILTRATION]: The skill only interacts with local SRT files and does not perform any network operations or unauthorized file access.
  • [COMMAND_EXECUTION]: The skill executes a bundled Python script (scripts/check_clean_srt.py) for validating subtitle quality. This script is local and does not use dangerous functions like eval or exec.
  • [REMOTE_CODE_EXECUTION]: All code logic is self-contained within the skill. No external packages or remote scripts are downloaded or executed at runtime.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided subtitle data but includes a validation script that enforces strict structural and content rules, minimizing the risk of processing malicious instructions.
  • Ingestion points: Reads original subtitle data from raw.srt as specified in SKILL.md.
  • Boundary markers: The process relies on the structured nature of the SRT file format.
  • Capability inventory: Uses the agent's ability to execute the provided validation script in the local environment.
  • Sanitization: The scripts/check_clean_srt.py script enforces character limits and removes punctuation, which limits the potential for external content to influence downstream agent logic.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 12, 2026, 04:56 PM