workflowlisted

# Workflow DAG For tasks with multiple steps and dependencies, submit a DAG to the server. The server-side WorkflowEngine handles scheduling, agent startup, failure policies, and completion notification. **Exit immediately after submission.** ## Environment Variables (Injected) | Variable | Description | |----------|-------------| | `EXPERT_API_BASE` | openteam-server HTTP address | | `OPENTEAM_CHAT_ID` | Current chat ID | | `OPENTEAM_INSTANCE_ID` | Your instance ID | ## Create Workflow ```bash bash {SKILL_DIR}/scripts/create-workflow.sh '<dag-json>' ``` DAG JSON format: ```json { "tasks": [ { "taskId": "t1", "agentId": "fullstack-engineer", "description": "Implement feature", "dependsOn": [], "onFailure": "stop" }, { "taskId": "t2", "agentId": "code-reviewer", "description": "Review implementation", "dependsOn": ["t1"], "onFailure": "stop" } ] } ``` Task fields: - `taskId`: unique identifier within the workflow - `agentId`: which expert agent to run this task - `description`: task description sent to the agent - `dependsOn`: array of taskIds that must complete first - `onFailure`: `stop` (halt DAG), `skip` (continue others), `retry` (re-run) - `maxRetries`: retry attempts (default 1, only for `retry` policy) - `maxRejects`: max times Lead can reject this task (default 2) - `timeoutMinutes`: per-task timeout (default 30) ## List Workflows ```bash bash {SKILL_DIR}/scripts/list-workflows.sh [status] ``` Filter by status: `running`, `suspended`, `completed`