`/prompts:speckit.clarify` with codex CLI offers recommended actions with no clarifying question

[colsonj]

Context:

I have established the project principles with /prompt:/speckit.constitution.
No edit's or changes have been requested of the originally generated markdown files.

I have then generated a baseline specification for example: /prompts:speckit.specify Define a delivery strategy for the project that slices features vertically

I have then used a prompt as an example below to ensure a minimum required clarifying questions are being asked:
/prompts:speckit.clarify DOMAIN & DATA MODEL — Generate the smallest set of questions to lock the core data contracts for the program (client-first). Ask for: (1) canonical field lists, types, and units for Contacts; (2) identifiers & keys and allowed character sets
...etc

I will then receive an output similar to below:

• Recommended: Option A – Explanation of the option

Option	Description
A	Do Step A.
B	Do Step B.
C	Do Step C.
Short	Provide a different short answer (<=5 words).

You can reply with the option letter (e.g., "A"), accept the recommendation by saying "yes" or "recommended", or provide your own short answer.

Issue
The output will only show the recommended steps, but does not give any context to what question is being asked

Expected Result

• DOMAIN & DATA MODEL - How will the canonical field lists, types, and units for Contacts be displayed?
• Recommended: Option A – Explanation of the option

Option	Description
A	Do Step A.
B	Do Step B.
C	Do Step C.
Short	Provide a different short answer (<=5 words).

You can reply with the option letter (e.g., "A"), accept the recommendation by saying "yes" or "recommended", or provide your own short answer.

Recommended Fix

• Update the generated .codex\prompts\speckit.clarify.md to include an additional line similar to below:

diff --git a\.codex\prompts\speckit.clarify.md b\.codex\prompts\speckit.clarify.md
--- a\.codex\prompts\speckit.clarify.md
+++ b\.codex\prompts\speckit.clarify.md
@@ -1,15 +1,16 @@
 4. Sequential questioning loop (interactive):
       - Present EXACTLY ONE question at a time.
       - For multiple-choice questions:
+          - Before any recommendation, restate the question verbatim on its own line as `**Question:** <original question>`.
             - **Analyze all options** and determine the **most suitable option** based on:
                 - Best practices for the project type
                 - Common patterns in similar implementations
                 - Risk reduction (security, performance, maintainability)
                 - Alignment with any explicit project goals or constraints visible in the spec
              - Present your **recommended option prominently** at the top with clear reasoning (1-2 sentences explaining why this is the best choice).
              - Format as: `**Recommended:** Option [X] - <reasoning>`
              - Then render all options as a Markdown table:
 
          | Option | Description |
          |--------|-------------|
          | A | <Option A description> |
          | B | <Option B description> |
          | C | <Option C description> (add D/E as needed up to 5) |
          | Short | Provide a different short answer (<=5 words) (Include only if free-form alternative is appropriate) |

Environment
Windows 11
Codex CLI - Run in WSL2 project virtual environment

[dmitry-pavlov]

Yeah I had it a couple of time as well

[MRiabov]

Same. Even on more powerful models.

[aurexav]

Sometimes the user input lost. I don't know why.

A tmp solution. Copy the command markdown template and replace the $ARGUMENTS part with your actual input.

[UrielCh]

https://www.calameo.com/books/004822953a07e84165740

[cmhac]

I see this issue in both copilot in vs code and codex via vscode extension. my solution has been to tweak the prompts to tell the agent to re-state the question before providing the suggested answers

[mnriem]

If you have a consistent reproduce with Copilot please add it to the issue for diagnosis