The Agent Skills Directory

[DATA_EXFILTRATION]: The skill facilitates reading a Notion API key from a configuration file at ~/.config/notion/api_key and sending it to an external service. While this is the intended functionality for an API tool and targets a well-known service (api.notion.com), it involves the handling of sensitive file paths and credential transmission. Evidence: NOTION_KEY=$(cat ~/.config/notion/api_key) and subsequent curl calls in SKILL.md.
[COMMAND_EXECUTION]: The skill utilizes shell commands via curl to perform its operations. Evidence: Multiple example curl commands in SKILL.md for searching, reading, and updating Notion pages.
[PROMPT_INJECTION]: The skill contains an indirect prompt injection surface because it retrieves and processes content from an external source (Notion). Ingestion points: Content retrieval from pages via GET /v1/blocks/{page_id}/children and search results. Boundary markers: Absent; there are no delimiters provided in the instructions to help the agent distinguish between data and potential instructions within the Notion content. Capability inventory: The skill can execute network requests and shell commands based on the data it retrieves. Sanitization: Absent; there is no mention of filtering or validating the content fetched from the API before the agent processes it.

notion