The Agent Skills Directory

[PROMPT_INJECTION]: The skill implements handlers that ingest untrusted external data from Slack messages and external incident reports, which are then interpolated into UI components.
Ingestion points: Slack message handlers (@app.message) and the incident dictionary in the build_notification_blocks function (SKILL.md).
Boundary markers: Absent. The skill does not use delimiters or warnings to isolate untrusted user content from the application's logic.
Capability inventory: The code uses the say() method and client.views_open() to transmit data back to Slack workspaces (SKILL.md).
Sanitization: While the code includes basic string truncation for the description field, it does not demonstrate escaping or sanitization of Markdown or control characters.
[SAFE]: The skill follows recommended security practices by using environment variables (os.environ) for managing sensitive credentials like SLACK_BOT_TOKEN and SLACK_SIGNING_SECRET.
[SAFE]: The code references established, well-known libraries from Slack (slack-bolt and slack-sdk) for its core functionality.

slack-bot-builder