The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's core functionality involves processing untrusted external data, which creates a surface for indirect prompt injection.
Ingestion points: The agent is instructed to fetch and read templates, shared documents via links, and data from integrations such as Slack, Teams, and Google Drive.
Boundary markers: The workflow does not include explicit instructions for the agent to use boundary delimiters or to ignore potential instructions embedded within the ingested data.
Capability inventory: The skill uses create_file and str_replace to manage the local file system and invokes sub-agents to test document content.
Sanitization: No methods for sanitizing or validating external content are specified before the agent drafts or refines document sections.
[COMMAND_EXECUTION]: The skill directs the agent to utilize filesystem tools, specifically create_file and str_replace, to create document scaffolds and perform iterative edits within the working directory.

doc-coauthoring