The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes system binaries for media processing.
Evidence: scripts/_mediaskills_common.py and scripts/to-captions-pipeline.py use subprocess.run to call ffmpeg and ffprobe. These calls use list-based arguments which mitigate shell injection risks.
[DYNAMIC_EXECUTION]: The test suite uses dynamic loading for module verification.
Evidence: tests/test_captions_compliance.py utilizes importlib.import_module to load and test individual script functions.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data which could contain malicious instructions.
Ingestion points: Multiple scripts, including validate.py, format.py, and to-scc.py, read SRT and media files from the file system provided via user-supplied paths.
Boundary markers: Absent. The skill does not implement delimiters or explicit 'ignore' instructions for the content of the processed files.
Capability inventory: The skill has the ability to execute shell commands (via ffmpeg/ffprobe) and write files to the workspace (.mediaskills/generated/).
Sanitization: While the skill performs standard SRT parsing and text normalization, it does not specifically sanitize input to prevent indirect prompt injection against the agent context.

captions-compliance