The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the nats CLI and curl to interact with NATS servers. These operations are encapsulated within helper functions (nats_cmd and nats_monitor) and are restricted by explicit safety rules that forbid state-changing operations (add, edit, delete, publish) without user consent.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data retrieved from NATS clusters, such as stream configurations and consumer statistics, creating a potential surface for indirect injection if cluster metadata is maliciously crafted.
Ingestion points: Data is ingested via nats CLI commands (e.g., stream list, consumer info) and curl requests to the monitoring API endpoints as defined in SKILL.md.
Boundary markers: No specific delimiters are used to wrap the output of the NATS commands.
Capability inventory: The skill is capable of executing shell commands through the nats binary and making network requests via curl to the configured NATS monitoring host.
Sanitization: The skill relies on jq to parse JSON responses and extract specific fields, which limits the data surface processed by the agent.

managing-nats