The Agent Skills Directory

[DYNAMIC_EXECUTION]: The script scripts/retention.py dynamically modifies the Python module search path (sys.path) using a computed relative path calculation. It specifically targets a sibling directory structure to import functionality.
Evidence: sys.path.insert(0, os.path.join(os.path.dirname(os.path.abspath(__file__)), "..", "..", "yt-edit", "scripts")) followed by from deadair import load as load_cues.
Impact: This allows the skill to load and execute code from arbitrary relative locations, which can be an attack vector if an attacker can control the contents of the yt-edit/scripts directory in the agent's environment.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external CSV and SRT (transcript) files provided by the user or fetched from remote sources.
Ingestion points: File paths retention.csv and transcript.srt are processed by scripts/retention.py.
Boundary markers: Absent. The script prints extracted data directly into the agent's context without delimiters or instructions to treat the content as data only.
Capability inventory: The skill enables the agent to read local files, execute Python scripts, and fetch remote data via yt-dlp.
Sanitization: The script performs basic float normalization for the CSV data and truncates transcript snippets to 140 characters, but does not sanitize potential natural language instructions embedded within the transcript.
[EXTERNAL_DOWNLOADS]: The skill provides instructions for the agent to use yt-dlp to fetch subtitle data from YouTube links provided during runtime.
Evidence: yt-dlp --skip-download --write-auto-subs --sub-langs ko --convert-subs srt "<video link>"
Note: This targets a well-known service (YouTube) and is documented neutrally as a functional capability for fetching transcripts.

yt-retention