main.js · part 200

60 text occurrences from desktop/Cursor.app/Contents/Resources/app/extensions/cursor-agent-host/dist/main.js, part 200. Every entry preserves the shipped literal and its saved verdict or selection reason.

File contents and all parts · All files

Video attachments require a video/* mime type, got ${p}

Attached media signed URLs are not available in this privacy mode

Video exceeds maximum size of ${v} bytes (${Math.round(v/1024/1024)}MB)

Video not found in blob store

Video exceeds maximum size of ${v} bytes (${Math.round(v/1024/1024)}MB)

Video attachment has no data

Video exceeds maximum size of ${v} bytes (${Math.round(v/1024/1024)}MB)

Failed to write video to filesystem

${a} still references an unresolved prompt upload.

Document not found in blob store

Failed to write document to filesystem

Skipping empty image attachment data

Failed to write long code selection to file

<code_selection lines="${s}-${o}" source="${r.path}" file="${t.path}">
The selection from `${r.path}` (lines ${s}-${o}) was too large to inline. Its full contents have been saved to `${t.path}`; read that file when you need the selection.
</code_selection>

<attached_files>
${N.map(e=>e.text).join("\n")}
</attached_files>

Folder: ${n}
Contents of directory:
${t.result}

<attached_folders>
Here are some folder${t.folders.length>1?"s":""} I manually attached to my message:

${e.join("\n\n")}
</attached_folders>

[${t.toLocaleString()} characters truncated due to size]

Truncated external link content before writing

Source URL: ${n}
Title: ${s}

Failed to write external link content to file

[${e.toLocaleString()} characters truncated due to size]

Blob store is required to hydrate selected pull request blob

Selected pull request blob not found

Blob store is required to hydrate git PR diff selection blob

Git PR diff selection blob not found

Blob store is required to hydrate extra context blob

Extra context blob not found

<external_links>
### Potentially Relevant Websearch Results

You should respond as if these information are known to you. Refrain from saying "I am unable to browse the internet" or "I don't have access to the internet" or "I'm unable to provide real-time news updates". This is your internet search results. Please always cite any links you referenced from the above search results in your response in markdown format.

-------

Website URL: ${t.url}
Website Title: ${t.title}
Website Content:
${t.content}
____

- ${e.path}

<uploaded_documents>
The following documents have been saved to your filesystem. You can read them using your file-reading tool or other tools:
${e}
</uploaded_documents>

- ${e.path}

 You can watch them using your WatchVideo subagent.

<attached_videos>
The following videos have been attached by the user and saved to your filesystem.${t}
${e}
</attached_videos>

<documentation_context>
## Potentially Relevant Documentation:
-------

Document Name: ${t.docName}
Document URL: ${t.pageUrl}
Document content:
${t.documentationChunk}
____

<cursor_rules_context>
Cursor Rules are extra documentation provided by the user to help the AI understand the codebase.
Use them if they seem useful to the users most recent query, but do not use them if they seem unrelated.

Rule Name: ${e.fullPath?function(e){const t=Math.max(e.lastIndexOf("/"),e.lastIndexOf("\\")),r=t>=0?e.substring(t+1):e;return r.toLowerCase().endsWith(".mdc")?r.substring(0,r.length-4):r}(e.fullPath):"Cursor Rule"}
Description: ${e.content?.slice(0,Tg)??""}

Cursor Rule

${e}
${r}
${t}

--- Cursor Command: ${e.name} ---
${e.content}
--- End Command ---

<cursor_commands>${e}
</cursor_commands>

  overview: ${t}

- ${e.name}${r}
  transcript_path: ${e.path}

<recent_agents_context>
The user has other recent agent conversations available as transcript files.
If they seem relevant to the user's current query, you may read them for additional context.
Do not read full transcript files in one go; search and read in targeted chunks.
${e}
</recent_agents_context>

<subagent_delegation_context>
The user has indicated they want you to delegate work to the following subagent(s): ${e}

To delegate, call the Task tool with the subagent_type parameter. Example:
Task(subagent_type="${t.selectedSubagents[0]?.name}", prompt="your detailed task description")
</subagent_delegation_context>

line ${e.startLine}

lines ${e.startLine}-${e.endLine}

<pr_review_content>
File: ${e.filePath}
Location: ${t}
PR URL: ${e.prUrl}

Diff Content:
${e.diffContent}

</pr_review_content>

<pr_review_context>
The user has attached diffs from a pull request
${e}
</pr_review_context>

${e.level} ${new Date(e.timestamp).toLocaleTimeString()}: ${e.message}

<console_logs_context>
Recent logs from the runtime connected to the AI agent.
${e}
</console_logs_context>

Invalid JSON

<ui_element>
React Component: ${e.component}
Props: ${t?JSON.stringify(t,null,2):"None"}
xpath: ${e.xpath}
</ui_element>

<ui_element>
${e.element}
xpath: ${e.xpath}
textContent: ${e.textContent}
</ui_element>

<selected_ui_elements>
The following UI elements have been selected by the user from the runtime.
${e}
</selected_ui_elements>

- ${e.pageTitle||"Untitled"}
  URL: ${e.url}
  Browser ID: ${e.browserId}