The Agent Skills Directory

[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection through the processing of external data sources.
Ingestion points: The skill reads meeting transcripts from Notion (pages prefixed with '@Date') and local context files within the .context/ directory.
Boundary markers: There are no defined delimiters or instructions to ignore embedded commands within the meeting transcripts being analyzed.
Capability inventory: The skill uses Read, Write, and Edit tools to update local files such as .context/current-focus.md and .context/projects/_index.md based on external input.
Sanitization: No sanitization or validation logic is present to filter malicious instructions embedded in meeting notes before they influence the agent's actions or local file modifications.

task-management