The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external news sources that could contain malicious instructions designed to manipulate the agent's output script.
Ingestion points: External news data fetched from sources like HackerNews or GitHub via the news-aggregator-skill dependency.
Boundary markers: No delimiters or warnings are used to isolate external content or instruct the agent to ignore instructions embedded within the news.
Capability inventory: The skill can execute local Python scripts (fetch_news.py, tts.py), run shell commands (ffmpeg), and write files to the local filesystem.
Sanitization: No evidence of input validation or sanitization for the fetched news items before they are incorporated into the podcast script.
[COMMAND_EXECUTION]: The skill orchestrates its workflow by executing local Python scripts and system binaries via the command line.
Evidence: Instructions in Step 2 call python3 to run a news fetching script located at skills/news-aggregator-skill/scripts/fetch_news.py.
Evidence: Instructions in Step 4 call python3 to run a TTS script at skills/tts/scripts/tts.py and invoke ffmpeg to concatenate generated audio files.
[EXTERNAL_DOWNLOADS]: The skill initiates network activity through its dependencies to fetch data and process audio.
Evidence: Fetches real-time news from public sources via the news-aggregator tool.
Evidence: Communicates with the author's official text-to-speech service at noiz.ai when the noiz backend is utilized for audio generation.

daily-news-caster