daemon.cjs · part 351

60 text occurrences from desktop/Cursor.app/Contents/Resources/app/extensions/cursor-agent-host/dist/agent-host-daemon/dist/bin/daemon.cjs, part 351. Every entry preserves the shipped literal and its saved verdict or selection reason.

File contents and all parts · All files

Unknown meta-parent task result

use strict

Use for forwarding relevant user-provided attachments to subagents. DO NOT use for non-user-provided files.

use strict

title is required

A short, user-friendly title for the agent to create. This appears in the UI as the agent's name. Make it concrete and distinct, consider recent titles to avoid reuse, and vary the first word across agent titles.

description is required

A brief description of this agent's scope of work. 10 words or less.

prompt is required

The prompt for the new asynchronous agent. DO NOT tell the agent that you are a meta-agent. Your prompt should be presented as just a normal user prompt.

responding_to_message_ids must contain at least one message id

Array of user message IDs this tool call is responding to. Usually this is the latest user message ID, but include older IDs for delayed actions or when a later message clarifies an earlier one. List multiple IDs when replying to multiple user messages.

fork must be a non-empty agent_id

Optional agent_id of a direct child agent to fork from. Use when you wish to start a new agent with the history of an existing agent. Common cases: parallel work that does not depend on the child's in-flight output, status check or question for a running child without interrupting its work, or multiple independent follow-ups from a completed child (e.g. splitting independent TODOs). Explicitly instruct new agent to fully shift its focus to the new task rather than continuing work on the prior task. Do not stop the work of the agent you are forking from (if you are doing this, resume that agent with `interrupt=true` instead). New agent must be given its own unique title.

new agent

Create an asynchronous agent with the provided prompt. Use `title` to provide a short user-facing name for the new agent. Returns agent_id of the created agent. Agent results will be reported asynchronously as user messages surrounded by 

use strict

Object expected.

Symbol.asyncDispose is not defined.

Symbol.dispose is not defined.

Object not disposable.

An error was suppressed during disposal.

Message must be non-empty

Message must be at most 100000 characters

`${target}` is not an agent link. Use `${id}`, `${id}#changes`, or `${id}#desktop`.

`${target}` is not an agent ID. A new agent's ID exists only in its CreateAgent result: link it after that result arrives, or name it without a link.

User-visible message to send.

Failed to send the message to the user: ${result.result.value.error}

Message sent to user.

Unhandled result case: ${String(exhaustiveCheck)}

Send a message to the user. This is the only channel the user sees; ordinary assistant text is hidden thinking. Send only when there is something the user needs — a result, a blocker, a question. Many turns (e.g. an irrelevant notification) end without any message. When you do send the turn's final message, make it the last thing you do.

Unknown error

use strict

agent_id is required

The agent_id returned by create-agent for the agent you want to message.

prompt is required

The follow-up prompt to send to the agent

responding_to_message_ids must contain at least one message id

Array of user message IDs this tool call is responding to. Usually this is the latest user message ID, but include older IDs for delayed actions or when a later message clarifies an earlier one. List multiple IDs when replying to multiple user messages.

Send a prompt to an existing asynchronous agent after it has completed. If the agent is still running, the request fails; wait for its completion before sending a follow-up.

Follow up for agent ${rawArgs.agent_id}

use strict

Object expected.

Symbol.asyncDispose is not defined.

Symbol.dispose is not defined.

Object not disposable.

An error was suppressed during disposal.

The content to display to the user, exactly as written.

Error: ${result.result.value.error}

Unhandled result case: ${String(exhaustiveCheck)}

Display a message directly to the user. Use this for progress updates, partial results, or content the user must see exactly as written before the task finishes.

Unknown error

Pi tool returned no result.

Unknown Pi tool error

Read the contents of a file using pi-compatible semantics.

Execute a bash command using pi-compatible output and timeout semantics.

Edit a single file using pi-compatible exact text replacements.

Write content to a file using pi-compatible semantics.

Search file contents for a pattern using pi-compatible grep semantics.

Search for files by glob pattern using pi-compatible find semantics.