The Agent Skills Directory

Potential for Indirect Prompt Injection: The skill summarizes communications from Teams, transcripts, and emails, which are untrusted external sources.
Ingestion points: The skill retrieves content through workiq_list_chat_messages, workiq_list_meeting_transcripts, and workiq_list_emails as specified in SKILL.md.
Boundary markers: The instructions do not specify the use of delimiters (like triple quotes or XML tags) to encapsulate the external content, which is a common technique to help the model distinguish between instructions and data.
Capability inventory: The skill is capable of writing configuration files to the local file system and creating tasks in Microsoft To Do using the workiq_create_task tool.
Sanitization: The instructions focus on summarizing the gist of the messages but do not explicitly require sanitization of the input to prevent the model from interpreting malicious instructions within the messages.
System Configuration via Command Execution: To ensure accurate scheduling, the skill uses a shell command (node -p) to identify the system's timezone. This demonstrates an interaction with the host environment that users should be aware of during the setup process.
Local Data Management: The skill stores user preferences and deduplication state in the .scout/ directory. While this is typical for maintaining state across automated runs, it involves standard file system access.

action-items-todo