The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides numerous curl and jq command snippets designed to interact with the Grafana Provisioning API. These commands utilize environment variables ($TOKEN, $GRAFANA, $UID) to manage authentication and target endpoints, which is a standard and safe practice for API tooling.
[INDIRECT_PROMPT_INJECTION]: The skill identifies a workflow where an agent reads alert rule data from a remote Grafana API, modifies it, and sends it back. This creates a potential surface for indirect prompt injection if alert fields (such as annotations or titles) were to contain instructions intended to influence the agent's behavior.
Ingestion points: Data retrieved from the /api/v1/provisioning/alert-rules/$UID and /api/v1/provisioning/folder/$FOLDER_UID/rule-groups/$GROUP endpoints via curl (documented in SKILL.md).
Boundary markers: The instructions do not define delimiters or specific "ignore" instructions for the ingested JSON content.
Capability inventory: The skill uses curl for both reading from and writing to the Grafana API, and jq for parsing the output.
Sanitization: There is no evidence of sanitization or validation of the rule content before it is processed or resubmitted.

grafana-alert-rule-api