Execute rauto directly for the user: run device commands, template execution, tx block, tx workflow, multi-device orchestration, replay, backup/restore, and...
--- name: rauto-usage description: "Execute rauto directly for the user: run device commands, template execution, tx block, tx workflow, multi-device orchestration, replay, backup/restore, and connection/profile/template/history operations. Trigger on requests like run command for me, execute on device, check history, run workflow, orchestrate multiple devices, apply staged rollout, use inventory/groups, diagnose profile, manage templates/profiles/connections, replay records, or restore backup." --- # Rauto Usage Execute rauto operations directly for users whenever possible. Do not default to tutorial-style answers. ## Core Mode Prefer action-first behavior: 1. Parse user goal and classify it as read-only or config-changing. 2. For read-only operations, run the relevant `rauto` command immediately. 3. For config-changing operations, prefer `tx`/`tx-workflow`/`orchestrate` with rollback-aware planning. 4. If commands are generated by the agent, ask for human confirmation before execution. 5. Return key results (not raw noise), plus command used. ## Execution Rules 1. For read/query requests, execute immediately: - examples: `device list`, `connection list`, `history list`, `templates list`, `replay --list`. 2. For config-changing requests, prefer rollback-capable execution: - single block: `tx` - multi-step/multi-resource: `tx-workflow` - multi-device staged rollout: `orchestrate` 3. For generated change commands, do not execute immediately: - show planned command(s), rollback strategy, and require user confirmation. 4. For explicit read-only commands (for example `show`), tx/workflow is not required. 5. Resolve connection in this priority: - explicit command args > `--connection <name>` > ask for missing fields. 6. Do not ask the user to manually run commands if agent can run them. 7. Summarize outputs with important fields: - target, mode, success/failure, key errors, next action. ## Risk Guardrails Require explicit user confirmation before destructive actions: - `rauto backup restore ... --replace` - profile/template/connection delete operations - tx/workflow execution that changes config when user intent is ambiguous Additional safety requirements for config changes: - Prefer `tx`/`tx-workflow`/`orchestrate` with rollback-aware planning over direct apply patterns. - Run/offer `--dry-run` first for workflow-based changes. - For `orchestrate`, review target scope, `fail_fast`, concurrency, and rollback boundary before execution. - Never auto-run agent-generated change commands without confirmation. - Include rollback path in the proposal before execution. If user explicitly asks to execute destructive action, proceed. ## Missing Input Strategy Ask only for missing must-have fields: - For `exec/template/tx/tx-workflow/orchestrate/connection test`: - need either full host credentials or usable `--connection`. - For `replay`: - need record file path or JSONL source. - For history queries: - need connection name. ## Response Format When command is executed, report: 1. `Operation`: what was run 2. `Command`: exact rauto command 3. `Result`: key output summary 4. `Notes`: risk, errors, or follow-up actions ## Navigation (Load References On Demand) - Agent execution decision tree and command templates: `references/agent-execution.md` - Full CLI command cookbook: `references/cli.md` - Runnable multi-device orchestration JSON templates (EN): `references/orchestration-json-template.md` - Multi-device orchestration pre-flight review: `references/orchestration-risk-check.md` - Runtime storage paths: `references/paths.md` - Troubleshooting and recovery: `references/troubleshooting.md` - End-to-end operation scenarios: `references/scenarios.md` - English Q/A examples: `references/examples.md` - Runnable workflow JSON templates (EN): `references/workflow-json-template.md` - Web tab/card mapping (only when user asks for Web operations): `references/web.md`
don't have the plugin yet? install it then click "run inline in claude" again.