The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes curl and jq via the Bash tool to query a VictoriaMetrics vmanomaly instance. It executes specific API calls to endpoints such as /health, /api/v1/server/buildinfo, and /api/v1/models to retrieve system status and model metadata for auditing purposes.
[DATA_EXFILTRATION]: The skill transmits the contents of local YAML configuration files to a user-provided $VM_ANOMALY_URL for validation via a POST request. While this involves sending local data over the network, it is a functional requirement of the auditing process. The skill explicitly instructs the user to redact credentials and use secure file permissions (mode-0600) for authentication headers.
[INDIRECT_PROMPT_INJECTION]: The skill ingests user-supplied YAML configuration files at runtime. These files provide the parameters and queries used by the agent to perform the audit. While this represents a data ingestion surface, the risk is mitigated by the skill's structured workflow and emphasis on manual evaluation of evidence.

vmanomaly-review