The Agent Skills Directory

[Indirect Prompt Injection] (HIGH): The skill demonstrates a high-risk surface for indirect prompt injection by aggregating data from external and untrusted sources into its decision-making and file-writing workflow.
Ingestion points: The skill reads data from 10_Daily/, 20_Project/, and 00_Inbox/ files, and crucially ingests external content from the /ai-newsletters and /ai-products workflows.
Boundary markers: Absent. There are no instructions to use delimiters or ignore instructions embedded within the newsletter digests or project notes.
Capability inventory: The agent has the capability to write and update files in 10_Daily/ and 00_Inbox/, and it uses the ingested data to 'Analyze & Prioritize' next steps.
Sanitization: Absent. External content is directly interpolated into the daily log and used to determine priorities without validation.
[Data Exposure] (LOW): The skill performs broad read operations across multiple directories including daily logs, project files, and inbox items. While consistent with the stated purpose of a daily planner, this level of access provides a significant target for data exposure if the agent is compromised via injection.

start-my-day