The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions direct the agent to use the curl utility to perform network operations for uploading image files.
Evidence: Both SKILL.md and references/notion-markdown.md specify using curl -sS -X POST "$UPLOAD_URL" -H "<header-name>: <header-value>" -F "file=@filename.png" for figure uploads.
Context: The destination URL and authorization headers are intended to be sourced from the output of the notion-create-file-upload tool. This is a functional requirement for file integration in the Notion environment.
[INDIRECT_PROMPT_INJECTION]: The skill demonstrates an attack surface for indirect prompt injection due to its broad data ingestion and significant write capabilities.
Ingestion points: Data is retrieved from existing Notion pages and templates via notion-fetch, and from external materials such as code, measurements, and meeting records mentioned in the "Normal Path" section of SKILL.md.
Boundary markers: The skill does not provide explicit instructions for the agent to use delimiters or to disregard instructions embedded within the source materials it processes.
Capability inventory: The skill utilizes notion-create-pages and notion-update-page for content modification, and shell access via curl for network communication.
Sanitization: Instructions in SKILL.md explicitly direct the agent to identify and exclude secrets, credentials, and personal data from the generated Notion pages.

write-notion