The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill provides a framework for the agent to ingest and process dashboard JSON and provisioning YAML files, which are used to interact with the Grafana API and manage server configurations. This surface could be exploited if malicious instructions are embedded within the configuration files provided by a user or external source.
Ingestion points: The agent is instructed to read and process dashboard JSON definitions and provisioning YAML configurations (e.g., provisioning/dashboards/default.yaml) as described in SKILL.md and references/dashboards.md.
Boundary markers: There are no explicit instructions or markers defined to delimit untrusted data or warn the agent to ignore embedded natural language instructions within the JSON/YAML content.
Capability inventory: The skill utilizes curl for interacting with the Grafana API (SKILL.md, references/api.md) and references the management of files in system paths such as /var/lib/grafana/ (SKILL.md).
Sanitization: No specific sanitization, schema validation, or escaping mechanisms are mentioned for the external content before it is processed or used in commands.

grafana-oss