The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's primary function involves ingesting untrusted data from external websites (blogs, social media, news) to generate summaries and briefings, which is a potential surface for indirect prompt injection attacks.
Ingestion points: Data is collected from company websites, LinkedIn, news articles, and review sites (G2, Capterra) via the research_pro agent as described in the Workflow and Step 3 sections of SKILL.md.
Boundary markers: The prompts provided in the column configuration (e.g., the News Summary prompt) do not explicitly define boundary markers or instructions to disregard embedded malicious commands within the fetched content.
Capability inventory: The skill has the capability to create and update tables via the extruct-api. It does not have direct access to the file system, shell, or sensitive credentials.
Sanitization: There are no defined sanitization or filtering steps for the content fetched from the web before it is passed to the LLM for synthesis in the news_summary column.

competitor-monitoring