The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses create_file and str_replace tools to manage documentation artifacts. It builds a file scaffold and performs surgical edits based on user feedback during the refinement stage.
[DATA_EXFILTRATION]: The workflow encourages users to provide extensive organizational context, including sensitive information such as team dynamics, past incidents, and politics, to improve document quality. It also facilitates connecting to well-known services like Slack, Google Drive, and SharePoint to retrieve context.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted data from multiple sources.
Ingestion points: External context is gathered via user 'info dumps', links to shared documents, and integration with messaging channels (Slack/Teams) and document storage.
Boundary markers: The instructions do not specify the use of delimiters or 'ignore embedded instructions' markers when processing the external content.
Capability inventory: The skill has the capability to write files (create_file), modify files (str_replace), and invoke sub-agents for reader testing.
Sanitization: There is no mention of sanitizing or validating the content pulled from external sources before it is drafted into the final document.

doc-coauthoring