The Agent Skills Directory

Data Access Patterns: This skill describes how to use specific tools to search through Slack data. It appropriately mentions that accessing private channels and direct messages requires user consent, which is a standard security measure to protect user privacy.
Indirect Prompt Injection Surface: The skill facilitates reading information from an external source (Slack). This creates a surface where the agent could potentially ingest instructions embedded in messages or files. This is a common characteristic of search-based skills.
Mandatory Evidence Chain for Indirect Injection: * Ingestion points: Slack search results (messages, files, channels). * Boundary markers: No specific boundary markers or instructions to ignore embedded commands are defined in this guidance. * Capability inventory: This file contains guidance only; no executable code or subprocess capabilities are implemented here. * Sanitization: No explicit sanitization or filtering of the retrieved content is discussed.

slack-search