The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructions specify the installation of prometheus-mcp-server via pip3 from PyPI. The source code is hosted on an unverified third-party GitHub repository (pab1it0/prometheus-mcp-server), which is not part of the trusted organizations list.
[COMMAND_EXECUTION]: To function, the skill executes the prometheus-mcp-server binary on the host system. It also relies on the availability of pip3 for the initial setup phase.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes data from external Prometheus endpoints (such as metric names, help text, and labels).
Ingestion points: Data enters the agent's context through list_metrics, get_metric_metadata, execute_query, and get_targets tools.
Boundary markers: There are no explicit delimiters or instructions provided to the agent to treat the retrieved Prometheus data as untrusted or to ignore embedded instructions.
Capability inventory: While the Prometheus tools themselves are read-only, the agent may possess other tools (e.g., file writing, network requests) that could be triggered by malicious strings found in metric labels or metadata.
Sanitization: The skill description does not mention any sanitization, filtering, or validation of the content returned by the Prometheus API.

prometheus-monitoring