The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from URLs, PDFs, YouTube videos, and podcasts, which presents a surface for indirect prompt injection attacks.
Ingestion points: URLs, PDF files, and media metadata/transcripts are ingested into the agent context.
Boundary markers: There are no explicit delimiters or instructions provided in the skill to isolate or ignore potentially malicious content within these sources.
Capability inventory: The skill references a local daemon and media tools including yt-dlp, ffmpeg, OCR, and transcription services.
Sanitization: The documentation does not specify any validation or sanitization steps for the data retrieved from external sources.
[EXTERNAL_DOWNLOADS]: The skill instructions direct the user to install third-party software from the npm registry and Homebrew.
Evidence: Instructions include npx -y @steipete/summarize, npm i -g @steipete/summarize, and brew install summarize.
[COMMAND_EXECUTION]: Setup and operation of the tool involve executing shell commands to manage a local background service and authentication tokens.
Evidence: Instructions describe running summarize daemon install --token <TOKEN> to pair the browser extension with a local background service on 127.0.0.1.

Summarize URLs, files, podcasts, and YouTube sources into agent-ready briefs with Summarize