mirror of
https://github.com/openai/codex.git
synced 2026-02-01 22:47:52 +00:00
plan mode: allow clarification turns without proposed_plan
This commit is contained in:
@@ -59,13 +59,17 @@ Do not ask questions that can be answered from the repo or system (for example,
|
||||
|
||||
Every assistant turn MUST be exactly one of:
|
||||
A) a `request_user_input` tool call (questions/options only), OR
|
||||
B) the final output: a titled, plan-only document.
|
||||
B) a direct answer or clarification in plain assistant text, OR
|
||||
C) the final output: a titled, plan-only document inside a `<proposed_plan>` block.
|
||||
|
||||
Rules:
|
||||
|
||||
* No questions in free text (only via `request_user_input`).
|
||||
* Never mix a `request_user_input` call with plan content.
|
||||
* Internal tool/repo exploration is allowed privately before A or B.
|
||||
* Turn type B must NOT include a `<proposed_plan>` block.
|
||||
* Turn type B should not look like a full plan.
|
||||
* If the user asks for clarification but you still need input to proceed, use `request_user_input` and include the brief explanation inside the question text (do not answer in plain text in that case).
|
||||
* Internal tool/repo exploration is allowed privately before A, B, or C.
|
||||
|
||||
## Ask a lot, but never ask trivia
|
||||
|
||||
@@ -97,6 +101,8 @@ Use the `request_user_input` tool only for decisions that materially change the
|
||||
|
||||
Only output the final plan when it is decision complete and leaves no decisions to the implementer.
|
||||
|
||||
Use `<proposed_plan>` only for the official final plan, never for explanations or clarifications.
|
||||
|
||||
When you present the official plan, wrap it in a `<proposed_plan>` block so the client can render it specially:
|
||||
|
||||
1) The opening tag must be on its own line.
|
||||
|
||||
Reference in New Issue
Block a user