The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill downloads Python dependencies (FastAPI, Slack SDK, etc.) from PyPI during the Slack notification setup process. These are standard, well-known libraries and are installed into a local, user-writable directory to comply with security constraints.
[COMMAND_EXECUTION]: The skill executes shell commands to interact with Docker, perform network probes via curl, and manage a local Python-based webhook server. All commands are scoped to the VSS deployment and the local environment.
[INDIRECT_PROMPT_INJECTION]: The skill processes natural language input via an agent endpoint and handles incident data from external webhooks. While this creates a potential surface for injection, the skill uses structured API interactions and the agent's internal logic serves as a buffer. The severity is low as it does not execute code from these inputs.
[SAFE]: The skill implements a mandatory two-step confirmation protocol for stopping or deleting alert rules, even in autonomous modes. This is a significant safety feature that prevents accidental or unauthorized deletion of monitoring rules.

vss-manage-alerts