rpk ai

Manage agents registered with the Redpanda AI platform. Agents are either managed (adp runs them; configured via a model, LLM provider, system prompt and MCP servers) or self-managed (a metadata-only record tracking a user-hosted agent).

Inspect or update agent containment.

Query agent relationships and provider failures.

Manage API keys (create, get, list, update, delete).

Read the ADP audit log: who did what, to which resource, from where, and which policy allowed or denied it, across the management API, the LLM proxy, the MCP gateway and agents. The log is read-only.

Manage rpk ai authentication (login, logout, token, status).

Manage Budgets (create, get, list, update, delete).

List and revoke the OAuth connections you hold to third-party providers. A connection is your personal grant to a provider (created by signing in through the consent flow) that lets user_oauth MCP servers act on your behalf.

Read published Agentic Data Plane guides and command references without signing in. With no path, browse a searchable picker in an interactive terminal, or print an index when piped.

Inspect the delivered authorization policy snapshot.

Manage rpk ai environments (list, use, add, show, rename, delete).

Manage Guardrails (create, get, list, update, delete).

Inspect telemetry ingestion diagnostics.

Install the Redpanda AI CLI. This command installs the latest version by default.

Manage Kill switch policies (create, get, list, update, delete).

Manage LLM providers (create, get, list, update, delete).

Manage MCP servers (create, get, list, update, delete).

Discover the models the Redpanda AI gateway exposes. The catalog is read-only; model list and model get are the only verbs.

Manage OAuth clients registered with the AI gateway’s OAuth Authorization Server. An OAuth client is an external tool (such as Claude AI, ChatGPT, or Cursor) that requests access tokens for an MCP server.

Manage OAuth authorization-server configurations registered with the Redpanda AI gateway. OAuth providers describe third-party authorization servers that user-facing MCP servers can authenticate against via the device-consent flow.

Manage policies and analyze or format draft Cedar.

Manage Policy templates (create, get, list, update, delete).

Launch a third-party coding agent configured to send its model traffic through the Redpanda AI gateway for the current environment. Authenticated sessions require an ADP API key.

Manage the secrets that LLM providers, MCP servers, OAuth clients, and agents reference by ID, for example with --openai-config.api-key-ref. Secret values are write-only: no command prints a stored value.

Inspect, rename or delete a managed agent’s conversation sessions.

Query spending and usage.

Inspect system policies and action groups.

Query traces, counts and tool sets.

Manage the triggers on an agent: the configured ways an agent is invoked from the outside (Microsoft Teams, or a cron schedule). A trigger belongs to exactly one agent, and its resource name carries that parent: agents/{agent}/triggers/{trigger}.

Uninstall the Redpanda AI CLI.

Upgrade to the latest Redpanda AI CLI version.

Print rpk ai version and commit.

string

Redpanda or rpk config file; default search paths are ~/.config/rpk/rpk.yaml, $PWD/redpanda.yaml, and /etc/redpanda/redpanda.yaml.

stringArray

Override rpk configuration settings; -X help for detail or -X list for terser detail.

bool

Ignore rpk.yaml and redpanda.yaml; use default settings.

string

rpk profile to use.

bool

Enable verbose logging.