The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill provides instructions for installing the umctl CLI tool using go install github.com/alibaba/UnifiedModel/cmd/umctl@latest. This is a vendor-provided resource belonging to the skill author.
[COMMAND_EXECUTION]: The skill facilitates the execution of the umctl CLI to perform entity, topology, and model queries, as well as curl and jq commands for executing telemetry query plans.
[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes JSON 'plans' from a UModel server which contain query strings and endpoint metadata used for subsequent execution steps. This creates an attack surface where a malicious or compromised server could return manipulated parameters to influence the agent's behavior.
Ingestion points: Output from umctl query run ... | entity-call get_metrics/get_logs as described in references/metrics-logs.md.
Boundary markers: None identified.
Capability inventory: The skill uses curl and shell execution to dispatch queries against external endpoints.
Sanitization: The skill provides instructions for manual endpoint overrides but lacks automated validation for the query content or endpoint targets received in the plan.
[DYNAMIC_EXECUTION]: In references/metrics-logs.md, the agent is instructed to extract PromQL or Elasticsearch DSL queries from a server-provided JSON object and interpolate them into shell commands for execution via curl.

umodel-query