菜单
Protocol commands for the Claude Code plugin: mirror runs into /workflows and notify the session. Run `smithers claude --help` for usage details.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Drive Smithers, a durable control plane for long-running coding agents. Use when the user wants multi-step, long-running, crash-safe, or human-in-the-loop agent work: "orchestrate agents", "run a workflow", "implement this and review it", "keep iterating until tests pass", "plan then build", or anything that needs retries, approvals, replay, or evals across multiple AI steps. YOU (the agent) run Smithers on the user's behalf; it is not a GUI the human clicks. You are an ORCHESTRATOR: run long-running, multi-step, or background work *through* Smithers, not through your own ad-hoc subagents; spend your time observing the run and reporting.
Drive Smithers, a durable control plane for long-running coding agents. Use when the user wants multi-step, long-running, crash-safe, or human-in-the-loop agent work: "orchestrate agents", "run a workflow", "implement this and review it", "keep iterating until tests pass", "plan then build", or anything that needs retries, approvals, replay, or evals across multiple AI steps. YOU (the agent) run Smithers on the user's behalf; it is not a GUI the human clicks. You are an ORCHESTRATOR: run long-running, multi-step, or background work *through* Smithers, not through your own ad-hoc subagents; spend your time observing the run and reporting.
Query node/run lifecycle history by default; pass --raw for raw agent chunks and all event types. Run `smithers events --help` for usage details.
Output detailed run state. Structured output canonically uses nodes[].nodeId (legacy steps[].id remains for compatibility); --pool tallies attempt engine/model usage. Run `smithers inspect --help` for usage details.
Build a new Smithers workflow from a plain-English description. Dispatches to the create-workflow builder. Run `smithers init` first if the .smithers/ pack is not yet installed. Run `smithers make-workflow --help` for usage details.
Concise run health at a glance: verdict, node counts, agent/model mix, throughput, and the nodes gating progress. Run `smithers status --help` for usage details.
| name | smithers-claude |
| description | Protocol commands for the Claude Code plugin: mirror runs into /workflows and notify the session. Run `smithers claude --help` for usage details. |
| requires_bin | smithers |
| command | smithers claude |
Follow the runs this session subscribed to (claude tick / claude subscribe) and print one NDJSON line per actionable transition (approval pending, human request, failed, stalled); --transitions all adds finished/cancelled/continued, --all-runs follows every run in the workspace. Backs the plugin's background monitor.
| Flag | Type | Default | Description |
|---|---|---|---|
--intervalMs | number | 2000 | Poll interval in ms |
--stalledAfterMs | number | 120000 | Heartbeat age that flags a running run as stalled |
--ticks | number | Stop after N polls (default: run until killed) | |
--transitions | string | actionable | Which transitions stream: actionable (approvals, human requests, failures, stalls) or all (also finished/cancelled/continued) |
--allRuns | boolean | false | Follow every run in the workspace instead of only the runs this session subscribed to (via claude tick / claude subscribe) |
Block until one node reaches a terminal state, then print its final state and output. Returns timedOut: true on --timeout-ms expiry (re-invoke to keep waiting).
| Name | Type | Required | Description |
|---|---|---|---|
nodeId | string | yes | Node ID to wait on |
| Flag | Type | Default | Description |
|---|---|---|---|
--runId | string | Run ID that owns the node | |
--iteration | number | Loop iteration (default: latest) | |
--timeoutMs | number | 480000 | Max wait in ms before returning timedOut: true |
--intervalMs | number | 1000 | Poll interval in ms |
--maxOutputChars | number | 2000 | Truncate the node output to this many chars |
Subscribe this session's background monitor to a run (done automatically by claude tick and Claude-launched runs); the monitor only notifies about subscribed runs.
| Name | Type | Required | Description |
|---|---|---|---|
runId | string | yes | Run ID the session's monitor should follow |
One /workflows mirror frame for a run: status, phase plan, nodes, deltas since --after-seq, outputs, approvals. --wait blocks until something relevant changes.
| Name | Type | Required | Description |
|---|---|---|---|
runId | string | yes | Run ID to mirror |
| Flag | Type | Default | Description |
|---|---|---|---|
--afterSeq | number | 0 | Event-log cursor from the previous tick's seq |
--wait | boolean | false | Block until a mirror-relevant event lands after --after-seq (or timeout) |
--timeoutMs | number | 420000 | Max wait in ms before returning timedOut: true |
--intervalMs | number | 750 | Wait poll interval in ms |
--maxOutputChars | number | 2000 | Truncate node outputs to this many chars |
--collapsePhases | boolean | false | Collapse the phase plan to a single phase |
Stop this session's background monitor from following a run (outside a Claude Code session it drops the run for every session).
| Name | Type | Required | Description |
|---|---|---|---|
runId | string | yes | Run ID the session's monitor should follow |
