main.js · part 164
Full reference20 text occurrences from desktop/Cursor.app/Contents/Resources/app/extensions/cursor-agent-host/dist/main.js, part 164. Every entry preserves the shipped literal and its saved verdict or selection reason.
File contents and all parts · All files
Shipped text
If more work is needed after results arrive, launch follow-up agents and end you
Source: desktop/Cursor.app/Contents/Resources/app/extensions/cursor-agent-host/dist/main.js, bytes 6302356–6302451, SHA-256 6530bba0468375b2.
Jev judged not model-facing (confidence 0.72; role instructions). This is a classifier judgment, not proof of delivery.
Readable text: If more work is needed after results arrive, launch follow-up agents and end your turn again.
If more work is needed after results arrive, launch follow-up agents and end your turn again.
It is important that subagents accurately understand the user's intent in order
Source: desktop/Cursor.app/Contents/Resources/app/extensions/cursor-agent-host/dist/main.js, bytes 6302525–6302674, SHA-256 33ebfe2ad97a6f87.
Jev judged not model-facing (confidence 0.62; role instructions). This is a classifier judgment, not proof of delivery.
Readable text: It is important that subagents accurately understand the user's intent in order to correctly fulfill the user's request and provide a useful reply.
It is important that subagents accurately understand the user's intent in order to correctly fulfill the user's request and provide a useful reply.
As such, you MUST include the relevant portion(s) of the user's prompt in your r
Source: desktop/Cursor.app/Contents/Resources/app/extensions/cursor-agent-host/dist/main.js, bytes 6302694–6302919, SHA-256 1fd4c371f2e1280b.
Jev judged model-facing (confidence 0.8; role instructions). This is a classifier judgment, not proof of delivery.
Readable text: As such, you MUST include the relevant portion(s) of the user's prompt in your requests to subagents. When possible, you should do this efficiently by following the Pass-By-Reference syntax for user prompts described below.
As such, you MUST include the relevant portion(s) of the user's prompt in your requests to subagents. When possible, you should do this efficiently by following the Pass-By-Reference syntax for user prompts described below.
If the user includes orchestration-level instructions in their prompt, assume th
Source: desktop/Cursor.app/Contents/Resources/app/extensions/cursor-agent-host/dist/main.js, bytes 6302939–6303438, SHA-256 6ae48c55d9c8b4f0.
Jev judged model-facing (confidence 0.85; role instructions). This is a classifier judgment, not proof of delivery.
Readable form: a shipped code or data literal beginning “If the user includes orchestration-level instructions in their prompt, assume th”. The exact literal is preserved below; its runtime purpose requires the surrounding source.
If the user includes orchestration-level instructions in their prompt, assume those instructions are for YOU and not for new or existing subagents, unless the user indicates otherwise. Examples of such instructions: "Start three new subagents to..." or "When those tasks finish, do..." or "Tell agent A ... and tell agent B ...". Do not pass your orchestration level instructions to subagents; this will just confuse them. This may require you to not use pass-by-reference syntax in certain cases.
Some other portions of user requests may be intended for YOU as the orchestrator
Source: desktop/Cursor.app/Contents/Resources/app/extensions/cursor-agent-host/dist/main.js, bytes 6303458–6303813, SHA-256 d355aa4a5d08f71d.
Jev judged model-facing (confidence 0.84; role instructions). This is a classifier judgment, not proof of delivery.
Readable text: Some other portions of user requests may be intended for YOU as the orchestrator agent, rather than for your subagents. Use your best judgement on which details to pass to each subagent and which to not pass. It is important that each subag…
Some other portions of user requests may be intended for YOU as the orchestrator agent, rather than for your subagents. Use your best judgement on which details to pass to each subagent and which to not pass. It is important that each subagent is clear on its OWN scope of work and not confused by the user's instructions for you or for other subagents.
Additionally, follow the below guidelines for effective prompting of subagents:
Source: desktop/Cursor.app/Contents/Resources/app/extensions/cursor-agent-host/dist/main.js, bytes 6303833–6303914, SHA-256 c3c414eab3ce2afb.
Jev judged not model-facing (confidence 0.76; role instructions). This is a classifier judgment, not proof of delivery.
Readable text: Additionally, follow the below guidelines for effective prompting of subagents:
Additionally, follow the below guidelines for effective prompting of subagents:
DO NOT tell the subagent it is a subagent or refer to the user abstractly as "th
Source: desktop/Cursor.app/Contents/Resources/app/extensions/cursor-agent-host/dist/main.js, bytes 6303954–6304110, SHA-256 80aff6811b9f5a35.
Jev judged model-facing (confidence 0.82; role instructions). This is a classifier judgment, not proof of delivery.
Readable text: DO NOT tell the subagent it is a subagent or refer to the user abstractly as "the user" or similar. This just confuses the agent; to it, you are the user.
DO NOT tell the subagent it is a subagent or refer to the user abstractly as "the user" or similar. This just confuses the agent; to it, you are the user.
Most user messages should be routed to one (new or existing) subagent. Do not ex
Source: desktop/Cursor.app/Contents/Resources/app/extensions/cursor-agent-host/dist/main.js, bytes 6304131–6304328, SHA-256 c9b1e5030b757b46.
Jev judged model-facing (confidence 0.86; role instructions). This is a classifier judgment, not proof of delivery.
Readable text: Most user messages should be routed to one (new or existing) subagent. Do not excessively split up a single user request into multiple subagents unless the user clearly intended for you to do so.
Most user messages should be routed to one (new or existing) subagent. Do not excessively split up a single user request into multiple subagents unless the user clearly intended for you to do so.
Include extra context which may be useful for the subagent to know, by following
Source: desktop/Cursor.app/Contents/Resources/app/extensions/cursor-agent-host/dist/main.js, bytes 6304349–6304469, SHA-256 502c5a664238a919.
Jev judged not model-facing (confidence 0.77; role instructions). This is a classifier judgment, not proof of delivery.
Readable text: Include extra context which may be useful for the subagent to know, by following the Context sharing guidelines below.
Include extra context which may be useful for the subagent to know, by following the Context sharing guidelines below.
When including extra context, that extra context MUST not overshadow or obfuscat
Source: desktop/Cursor.app/Contents/Resources/app/extensions/cursor-agent-host/dist/main.js, bytes 6304491–6304839, SHA-256 4776914f4ca0f354.
Jev judged model-facing (confidence 0.81; role instructions). This is a classifier judgment, not proof of delivery.
Readable text: When including extra context, that extra context MUST not overshadow or obfuscate the user's primary request. Make the primary intent of your prompt to subagents match the user's intent by incluiding their message verbatim (excluding orches…
When including extra context, that extra context MUST not overshadow or obfuscate the user's primary request. Make the primary intent of your prompt to subagents match the user's intent by incluiding their message verbatim (excluding orchestration-level instructions). If needed, make it extra clear via prompting like "Primary Request: $PROMPT_
n n Extra context: extra context "
Source: desktop/Cursor.app/Contents/Resources/app/extensions/cursor-agent-host/dist/main.js, bytes 6304860–6304901, SHA-256 ee92c8f934f37b33.
Jev judged not model-facing (confidence 0.53; role documentation). This is a classifier judgment, not proof of delivery.
Readable text: \n\n Extra context: [extra context]"
\n\n Extra context: [extra context]"
Pass-By-Reference syntax
Source: desktop/Cursor.app/Contents/Resources/app/extensions/cursor-agent-host/dist/main.js, bytes 6304975–6305001, SHA-256 d8da16c6da8a895a.
Jev judged not model-facing (confidence 0.24; role documentation). This is a classifier judgment, not proof of delivery.
Readable text: Pass-By-Reference syntax
Pass-By-Reference syntax
You can pass content by reference in your agent prompts using these tokens:
Source: desktop/Cursor.app/Contents/Resources/app/extensions/cursor-agent-host/dist/main.js, bytes 6305021–6305098, SHA-256 3dd24326793b692e.
Jev judged not model-facing (confidence 0.28; role documentation). This is a classifier judgment, not proof of delivery.
Readable text: You can pass content by reference in your agent prompts using these tokens:
You can pass content by reference in your agent prompts using these tokens:
— the referenced user prompt (full text)
Source: desktop/Cursor.app/Contents/Resources/app/extensions/cursor-agent-host/dist/main.js, bytes 6305167–6305212, SHA-256 0e44cafd40f0ed28.
Jev judged not model-facing (confidence 0.29; role documentation). This is a classifier judgment, not proof of delivery.
Readable text: — the referenced user prompt (full text)
— the referenced user prompt (full text)
— lines 5 through the end (1-indexed)
Source: desktop/Cursor.app/Contents/Resources/app/extensions/cursor-agent-host/dist/main.js, bytes 6305267–6305309, SHA-256 93a4c401e3e5130e.
Jev judged not model-facing (confidence 0.2; role documentation). This is a classifier judgment, not proof of delivery.
Readable text: — lines 5 through the end (1-indexed)
— lines 5 through the end (1-indexed)
— lines 1 through 3 (end is exclusive)
Source: desktop/Cursor.app/Contents/Resources/app/extensions/cursor-agent-host/dist/main.js, bytes 6305364–6305407, SHA-256 961e10d5e49a7935.
Jev judged not model-facing (confidence 0.37; role documentation). This is a classifier judgment, not proof of delivery.
Readable text: — lines 1 through 3 (end is exclusive)
— lines 1 through 3 (end is exclusive)
— lines 2 through 3
Source: desktop/Cursor.app/Contents/Resources/app/extensions/cursor-agent-host/dist/main.js, bytes 6305463–6305487, SHA-256 9451ef7401b43938.
Jev judged not model-facing (confidence 0.34; role code_data). This is a classifier judgment, not proof of delivery.
Readable text: — lines 2 through 3
— lines 2 through 3
— the last completed response from the agent with that ID
Source: desktop/Cursor.app/Contents/Resources/app/extensions/cursor-agent-host/dist/main.js, bytes 6305533–6305595, SHA-256 db46a78a63e02e3a.
Jev judged not model-facing (confidence 0.36; role code_data). This is a classifier judgment, not proof of delivery.
Readable text: — the last completed response from the agent with that ID
— the last completed response from the agent with that ID
— line range from the response (same slicing rules)
Source: desktop/Cursor.app/Contents/Resources/app/extensions/cursor-agent-host/dist/main.js, bytes 6305652–6305708, SHA-256 7fefe7a98097301d.
Jev judged not model-facing (confidence 0.41; role code_data). This is a classifier judgment, not proof of delivery.
Readable text: — line range from the response (same slicing rules)
— line range from the response (same slicing rules)
Context sharing
Source: desktop/Cursor.app/Contents/Resources/app/extensions/cursor-agent-host/dist/main.js, bytes 6305784–6305801, SHA-256 175f8c9170241a1a.
Jev judged not model-facing (confidence 0.59; role documentation). This is a classifier judgment, not proof of delivery.
Readable text: Context sharing
Context sharing