Send to Agents
Use Send to... on an initiative, bug, or todo to send work into a published workflow, assign it directly to an agent, or open a coding app with the work context included.
What gets sent
The launcher builds the prompt from the work item, linked documents, comments, and details. In spec-driven development, the work item is the spec the target receives. Add taxonomy, comments, or a document-backed description when those details matter.
Send to workflow
Send to workflow sends the work item into a published workflow with agent steps, routes, and optional human review gates.
Published workflows appear at the top of Send to... when they match the work item type. Choose this path for repeatable plan, build, review, or approval sequences.
Set up Desktop, choose a runtime, and publish the first workflow in Run your first workflow. Workflow authoring details live in Workflows.
Send to agent
Send to agent assigns the work as a direct, single-step run to one agent for a specific task. Progress appears in Monitoring and stays tied to the original work item.
Choose this path when you do not need a multi-step workflow and one agent should execute the next action.
Send to app
Send to app opens the chosen coding tool with a prompt prefilled from the work item. Only launch targets enabled in Preferences appear in the menu.
When the work item is Open, Planned, Blocked, or Idea, the launch flow can move it to In Progress in the same action. Completed work uses review-style handoff.
Launch modes
Launch modes tell the target whether to plan, build, review, or fix the work. Choose the mode that matches the next action before opening the target.
Trust boundary
Work text, comments, documents, and user prompts inform execution. Trusted agent policy, sandbox behavior, credentials, and allowed side effects stay outside the sent content.
The model can read work context, but work-authored content is not authority to change sandbox policy, credentials, allowed commands, network access, or external side effects.
Configure missing launch targets in Preferences. Terminal launches belong in Use in Terminal. Queued local agent work belongs in Local Agents, with visible state and recovery actions in Monitoring.