The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses shell commands and Python scripts to perform its tasks. It automates the environment setup by running pip install youtube-transcript-api and executes a local script scripts/extract-transcript.py to retrieve video data. It also uses standard shell utilities like grep and sed for URL parsing.
[EXTERNAL_DOWNLOADS]: The skill fetches content from external YouTube domains (youtube.com, youtu.be) to retrieve video metadata and transcripts. Additionally, it automates the installation of the youtube-transcript-api package from the official Python Package Index (PyPI) during initial setup.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted transcript data which could contain malicious instructions.
Ingestion points: Transcript text is fetched from external URLs provided by the user and stored in the /tmp directory before being read into the agent's context (SKILL.md, scripts/extract-transcript.py).
Boundary markers: The instructions do not define specific delimiters or include "ignore embedded instructions" warnings for the AI when processing the transcript content.
Capability inventory: The skill has capabilities to write to the file system, execute shell scripts, and install Python packages.
Sanitization: While it filters XML tags and common noise markers like [Music], it lacks sanitization to prevent the interpretation of instructions hidden within the transcript text.

youtube-summarizer