Bug Report: TUI prompts appear during `copier update` despite `--defaults --skip-answered` flags

[SteelOtter1242]

Describe the problem

Bug Report: TUI prompts appear during copier update despite --defaults --skip-answered flags

Description

When running copier update with --defaults --skip-answered flags in automation contexts (Make targets, shell scripts), the interactive TUI still appears and prompts for questions that have already been answered. This breaks non-interactive workflows and causes commands to hang waiting for user input.

Expected Behavior

When using --defaults --skip-answered flags, copier should:

1. Use default values for all questions
2. Skip questions that already have answers in .copier-answers.yml
3. Run completely non-interactively without showing the TUI
4. Complete successfully without requiring terminal input

Actual Behavior

Despite using --defaults --skip-answered flags, copier:

1. Shows the interactive TUI with emoji prompts (🎤)
2. Displays questions that already have answers
3. Waits for user input, causing the command to hang
4. Prevents automation/CI/CD workflows from completing

Reproduction Steps

1. Create a template and generate a project with copier
2. Ensure project has .copier-answers.yml with previous answers
3. Run update from a Makefile or script using $() command substitution:

   update:
       OUTPUT=$$(copier update --trust --defaults --skip-answered --vcs-ref=HEAD 2>&1); \
       echo "$$OUTPUT"

4. The command hangs showing interactive prompts despite flags

Environment

• Copier version: 9.11.0
• Python version: 3.12
• OS: Linux
• Context: Make target with $$() command substitution

Root Cause

Copier detects if stdin is connected to a terminal and launches the TUI even when --defaults --skip-answered flags are present. In Make's $$() subprocess, stdin remains open/connected, causing copier to attempt interactive mode.

Workaround

Close stdin explicitly when running copier in a subshell:

( copier update --trust --defaults --skip-answered --vcs-ref=HEAD 0<&- < /dev/null ) > output.txt 2>&1

Or in a Makefile:

OUTPUT=$$($$COPIER_CMD update $$OPTS 0<&- < /dev/null 2>&1);

This forces copier into non-interactive mode by closing file descriptor 0.

Impact

This issue affects:

• ✗ Makefile-based automation
• ✗ CI/CD pipelines using shell script wrappers
• ✗ Any subprocess that captures output via $()
• ✗ Automated template updates in frameworks

Suggested Fix

The --defaults and --skip-answered flags should take precedence over stdin detection. If these flags are present, copier should:

1. Never attempt to show the TUI
2. Raise an error if a required question has no default/answer (as documented)
3. Run in truly non-interactive mode regardless of stdin state

Alternatively, document that stdin must be explicitly closed when running copier in subprocess contexts for automation.

References

• Existing issue that added --skip-answered: Don't require users to review all questions when running copier update #1178
• Documentation: https://copier.readthedocs.io/en/stable/configuring/#skip_answered
• Test coverage exists but uses pexpect.spawn() with PTY, not subprocess $()

Additional Context

The extensive test suite for --skip-answered uses spawn() which creates a pseudo-terminal (PTY). This is different from Make's $$() subprocess behavior where stdin is inherited from the parent process. The tests pass because PTY behaves differently than real subprocess stdin redirection.

The workaround is reliable but non-obvious. Many users expect --defaults --skip-answered to mean "never prompt, use existing answers" without needing to understand stdin file descriptor manipulation.

Template

Digitalized Architecture Framework - Template Configuration

=============================================================

_metadata:
version: "1.1.0"
description: "Digitalized Architecture Framework with Four Architecting Pillars"

Default answers (for --defaults mode)

project_name: "my-project"
project_description: "A software development project using the Digitalized Architecture Framework"
organization_name: "Development Team"
methodology_folder: "daf-methodology"
workspace_folder: "daf-workspace"
tests_folder: "tests"
use_capability_areas: true
use_user_journeys: true
use_issue_templates: true
use_test_methodology: true
use_reports: false

Questions for interactive mode

_questions:
project_name:
type: str
help: "What is the name of your project?"
placeholder: "my-awesome-project"
default: "my-project"

project_description:
type: str
help: "Brief description of your project"
placeholder: "An innovative solution for..."
default: "A software development project"

organization_name:
type: str
help: "Your organization name"
placeholder: "Acme Corp"
default: "Development Team"

workspace_folder:
type: str
help: "Name for the workspace folder (your project work)"
placeholder: "daf-workspace, ai-project, my-workspace"
default: "daf-workspace"

tests_folder:
type: str
help: "Name for the tests folder (test suite location)"
placeholder: "tests, test, tst"
default: "tests"

use_capability_areas:
type: bool
help: "Include Capability Area methodology?"
default: true

use_user_journeys:
type: bool
help: "Include User Journey methodology?"
default: true

use_issue_templates:
type: bool
help: "Include issue tracking templates?"
default: true

use_test_methodology:
type: bool
help: "Include testing methodology?"
default: true

use_reports:
type: bool
help: "Include reporting tools and templates?"
default: false

Template processing

_templates_suffix: ".jinja2"
_envops:
keep_trailing_newline: true

Skip existing files by default (non-destructive installation)

_skip_if_exists:

• "README.md"
• "*/README.md"

Files to exclude

_exclude:

• "copier.yml"
• "copier.yml.bak"
• ".copier-answers.yml" # Framework's answers (adopters generate their own)
• ".git"
• ".gitignore"
• "*.pyc"
• "pycache"
• ".DS_Store"

Framework development (not shipped to users)

• "daf-workspace"
• "daf-workspace/**"
• "pytest.ini" # Dev version includes daf-workspace/tests
• "install.sh"
• "pdf"
• "pdf/**"
• "tests"
• "tests/**"

Install scripts and templates (used by install.sh, not copied)

• "daf-methodology/bin/install"
• "daf-methodology/bin/install/**"

Symlinked directories (do not copy symlinks, copy actual source files instead)

• "daf-methodology/docs/_root"
• "daf-methodology/docs/_root/**"

Generated files

• "**/00-OVERVIEW.md" # Generated by navigation script
• "daf-methodology/docs/index.md" # Symlink (generated)

Framework-specific files (not for adopters)

• "mkdocs-ai-dev.yml" # Framework dev only (not copied to adopters)
• "CONTRIBUTING.md" # Copied as DAF-CONTRIBUTING.md by install.sh

Tasks to run after generation

_tasks:

Note: Copier automatically creates .copier-answers.yml for update support

• "echo '✅ Digitalized Architecture Framework installed successfully!'"
• "echo ''"
• "echo '📊 Framework: 8 Methodologies across 4 Pillars'"
• "echo ' Pillar 1: Value Stream Architecting (User Journey, Domain Modeling)'"
• "echo ' Pillar 2: Complexity Management Architecting (Capability Area, Arc42++)'"
• "echo ' Pillar 3: Product Memory Architecting (Knowledge Map, Conversation Backlog)'"
• "echo ' Pillar 4: Quality Assurance Architecting (Testing Pyramid, Delivery Modeling)'"
• "echo ''"
• "echo '📖 What was created:'"
• "echo ' • docs/knowledge-map.html - Interactive methodology visualization'"
• "echo ' • README.md - Framework overview with Four Pillars'"
• "echo ' • daf-methodology/make/backlog.mk - Makefile extension for backlog management'"
• "echo ' • daf-methodology/bin/backlog/ - Backlog management scripts (CLI, TUI)'"
• "echo ' • .github/instructions/ - AI agent instruction files (context-optimized)'"
• "echo ' • .github/prompts/ - README enhancement prompts for AI agents'"
• "echo ' • docs/capability-areas/ - Feature development templates'"
• "echo ' • docs/user-journeys/ - User experience templates'"
• "echo ' • issues/templates/ - Issue tracking templates (Conversation Backlog)'"
• "echo ' • tests/ - Testing Pyramid methodology'"
• "{% if use_reports %}echo ' • reports/ - Reporting tools'{% endif %}"
• "echo ''"
• "echo '📚 Quick Start:'"
• "echo ' 1. Explore: Run make meth-serve (Ctrl+Click URL to open docs)'"
• "echo ' 2. Document: Start in daf-workspace/docs/capability-areas/'"
• "echo ' 3. Issues: Run make issues-ui (interactive TUI)'"
• "echo ''"
• "echo '⚙️ Optional (custom folder names): Edit daf-config.yml, then run make config-update'"
• "echo '📦 Updates: Run make daf-update to get latest framework'"

To Reproduce

No response

Logs

Expected behavior

it should work without TUI

Screenshots/screencasts/logs

No response

Operating system

Linux

Operating system distribution and version

Ubuntu 24

Copier version

Copier 9.11.0

Python version

Python 3.12.3

Installation method

uvx+pypi

Additional context

No response

[sisp]

I'm afraid I can't reproduce the problem you're describing, @SteelOtter1242.

Here is the minimal example I've used: example.zip

And these are my commands to try reproducing the problem:

$ uvx copier copy -r v1 src dst
🎤 name
   john

Copying from template version 1
    create  test.txt
    create  .copier-answers.yml

$ cd dst
$ git init
$ git add .
$ git commit -m 'v1'
$ cd ..
$ make update
OUTPUT=$(uvx copier update --trust --defaults --skip-answered --vcs-ref=HEAD dst 2>&1); \
echo "$OUTPUT"
Updating to template version 2

$ cd dst/
$ git status
On branch main
Changes not staged for commit:
  (use "git add <file>..." to update what will be committed)
  (use "git restore <file>..." to discard changes in working directory)
	modified:   .copier-answers.yml
	modified:   test.txt

no changes added to commit (use "git add" and/or "git commit -a")

$ git diff
diff --git a/.copier-answers.yml b/.copier-answers.yml
index 05e181f..dfd5a64 100644
--- a/.copier-answers.yml
+++ b/.copier-answers.yml
@@ -1,3 +1,3 @@
-_commit: v1
+_commit: v2
 _src_path: src
 name: john
diff --git a/test.txt b/test.txt
index 91b2aaa..dc0d10b 100644
--- a/test.txt
+++ b/test.txt
@@ -1 +1,2 @@
 name: john
+edited

---

Off-topic question: Is the content under "Template" AI-generated? It looks quite verbose and not closely related to the problem you've described. A minimal reproducible example is much more helpful, so we can more easily reproduce and narrow down the problem.