The Agent Skills Directory

[DATA_EXFILTRATION]: The skill performs network operations using curl to https://ai-factory.frevana.com in scripts/search_google_news.sh. This transmission includes the user's search query (q) and a Bearer token (FREVANA_TOKEN) provided in the environment or as an argument. While this is the intended purpose of the skill, it involves sending potentially sensitive search context to a non-whitelisted third-party domain.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the Google News API.
Ingestion points: External news content is fetched via curl and stored in RESPONSE_FILE within scripts/search_google_news.sh.
Boundary markers: None are present; the skill returns the raw JSON response or a summary of its contents.
Capability inventory: The skill has file-write capabilities (--output flag) and network access (curl).
Sanitization: The script performs JSON validation but does not sanitize the text content of news articles. If the agent subsequently processes this content as instructions, it could be influenced by malicious content embedded in news snippets.
[COMMAND_EXECUTION]: The skill executes bash and python3 to facilitate its core logic. It allows the user to specify an --output path, which is used in scripts/search_google_news.sh with mkdir -p and cp. This allows the skill to write files to any path accessible by the executing user, which could be abused to overwrite configuration files if not carefully managed by the agent.
[DYNAMIC_EXECUTION]: The bash script scripts/search_google_news.sh uses python3 - with heredoc blocks to dynamically execute Python code for JSON processing. While the embedded code is static and used for data formatting and validation, it represents a dynamic execution pattern.

google-news-search