The Agent Skills Directory

Indirect Prompt Injection Surface: The skill ingests untrusted data from Microsoft Planner (task titles, descriptions, and checklists) via the fetch_work_iq tool and incorporates this content into generated reports. If a task contains malicious instructions, they could potentially influence the agent's behavior during report generation or affect the user's perception.
Ingestion points: Task metadata and details are fetched in Step 4 from /planner/plans/{planId}/tasks and /planner/tasks/{taskId}/details.
Boundary markers: The instructions do not specify using delimiters (like XML tags or triple quotes) to isolate external task content from the report instructions.
Capability inventory: The skill utilizes create_entity_work_iq to send reports via email or Teams and fetch_work_iq for data retrieval.
Sanitization: No explicit sanitization or filtering of task descriptions is defined before they are placed into the report Markdown.
Outbound Data Protection: The skill includes a clear instruction to preview the report and obtain explicit user confirmation before using create_entity_work_iq to share content externally (Step 9). This is a positive security practice that mitigates unauthorized data exfiltration.
Rendering Controls: The skill implements a 'Renderer capability policy' (Step 8) to safely handle Mermaid chart generation. It checks for host-level capability support and respects user opt-outs, preventing issues related to unsupported or unwanted visual rendering.

planner-status-report