The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes shell commands such as curl and jq to interact with the Notion API and process JSON data. These operations are restricted to the core functionality of publishing meeting records.
[EXTERNAL_DOWNLOADS]: Communicates with api.notion.com to query databases and create pages. Notion is a well-known and established service, and the use of its official API is appropriate for the skill's stated purpose.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests and processes untrusted data from meeting transcripts and external files.
Ingestion points: Meeting content is ingested from user-pasted text (Source A) or read from file paths provided by the user (Source B) in SKILL.md.
Boundary markers: The skill does not implement delimiters or explicit instructions to the agent to ignore embedded commands within the meeting content.
Capability inventory: The skill possesses network access via curl and file-reading capabilities for local configuration in SKILL.md.
Sanitization: The skill employs jq --arg when building the final JSON payload in Phase 4-3, which effectively escapes the untrusted meeting content and prevents traditional shell command injection.

write-meeting-notes