The Agent Skills Directory

[COMMAND_EXECUTION]: The skill documentation provides standard installation commands for required Motoko packages (mops add slack-client@0.1.0 and mops add caffeineai-authorization@1.0.1). These are legitimate dependency management operations.
[EXTERNAL_DOWNLOADS]: The skill fetches dependencies from the Mops package registry. All external links point to official Slack documentation (api.slack.com) or the Mops registry (mops.one), which are trusted sources for this context.
[CREDENTIALS_SAFE]: The skill handles Slack API tokens (xoxb-, xoxp-) securely. It explicitly instructs the developer to gate token storage behind an admin permission check (AccessControl.hasPermission(..., #admin)), ensures tokens are never returned to the frontend, and passes them via Authorization: Bearer headers rather than URL parameters to prevent logging leaks.
[DATA_EXFILTRATION]: While the skill communicates with Slack, this is the core intended purpose. The implementation uses non-replicated outcalls, ensuring sensitive headers are handled by individual nodes rather than the entire subnet during consensus, which is a standard security practice for IC outcalls.

connector-slack