The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external news feeds which could contain malicious instructions designed to influence the agent's behavior during summarization.
Ingestion points: Fetches multiple JSON files (daily-brief.json, latest-24h.json, top3-personas.json, stories-merged.json, source-status.json) from learnprompt.github.io in SKILL.md.
Boundary markers: Absent. The skill does not use specific delimiters or instructions to the agent to ignore embedded commands within the fetched news items.
Capability inventory: The skill environment allows for shell command execution (curl) and Python 3 processing.
Sanitization: Absent. The embedded Python scripts in SKILL.md extract and print raw string values (titles, reviews, source names) from the JSON payloads without filtering or escaping content.
[COMMAND_EXECUTION]: The skill relies on the agent executing shell commands and Python scripts to fetch and filter data.
Evidence: Multiple code blocks in SKILL.md use curl to download data and python3 heredocs to process the results (e.g., lines 72, 85, 96, 105).
[EXTERNAL_DOWNLOADS]: The skill downloads static JSON data from external web sources.
Evidence: Requests target https://learnprompt.github.io/ai-news-radar/data/ and https://raw.githubusercontent.com/LearnPrompt/ai-news-radar/master/data/. These resources are owned by the vendor and are used for legitimate news delivery.

ai-radar