Agent skills
Sume CLI commands are designed to make agent automation explicit and auditable.
Use read-only first
Prefer read commands while planning:
Require confirmation for writes
Agents should not create resources or submit paid provider work without explicit operator confirmation.
| Gate | Use for |
|---|---|
--confirm-submit | Non-paid writes such as job cancellation or asset registration. |
--confirm-paid | Generation that can reserve or spend credits (Avatar / Avatar Video). |
Image, Video, and Music generation are not CLI submit commands yet — agents should call the Developer API for those families and still use CLI job recovery where helpful.
Redact sensitive output
Use --agent --json when automation reads command output. Agent mode redacts or
summarizes URL-like fields and account/workspace details where supported.
Bundled skills
Install or refresh the packaged Sume skill into local agent skill directories:
sume skills install writes into .agents/skills or .claude/skills. Use
export to review source files before a custom install.
MCP for agents
Prefer hosted MCP (https://mcp.sume.com/mcp) for Cursor/Claude remote
connectors. Local sume mcp is not launched in current CLI releases — see
sume mcp doctor --json and the CLI command reference.
Recommended loop
- Inspect catalog and local readiness.
- Validate the payload with a read-only/schema command.
- Ask for confirmation before writes or paid work.
- Submit one bounded job (CLI for Avatar; API for Image/Video/Music).
- Recover through job status/events/result commands.
- Summarize outputs without pasting secrets or signed URLs.