The Agent Skills Directory

Processing of Untrusted Message Content: The skill retrieves and analyzes summaries of recent messages within Teams channels to determine activity levels. Because these messages are authored by various users, they represent untrusted data that could theoretically contain instructions intended to influence the agent's summary or health classification.
Evidence: SKILL.md Step 3 specifically requests a 'content summary' for the 10 most recent messages in each channel.
Consideration: This is a common pattern for audit tools. Since the final output is a report presented to the user for review, the risk is limited to potential inaccuracies in the generated audit summary if a message contains deceptive content.

channel-audit