feat: Python env paths redesign v2 with PowerShell fixes

.gitignore

@@ -85,6 +85,7 @@ venv/
 ENV/
 env/
 .conda/
+.envs/
 
 # Testing
 .pytest_cache/
@@ -168,4 +169,5 @@ OPUS_ANALYSIS_AND_IDEAS.md
 
 # Auto Claude generated files
 .security-key
-/shared_docs
+/shared_docs
+tmpclaude*

.husky/pre-commit

@@ -131,7 +131,7 @@ if git diff --cached --name-only | grep -q "^apps/backend/.*\.py$"; then
       echo "$STAGED_PY_FILES" | xargs git add
     fi
   else
-    echo "Warning: ruff not found, skipping Python linting. Install with: uv pip install ruff"
+    echo "⚠️  Warning: ruff not found, skipping Python linting. Install with: uv pip install ruff"
   fi
 
   # Run pytest (skip slow/integration tests and Windows-incompatible tests for pre-commit speed)

CLAUDE.md

@@ -14,10 +14,11 @@ Auto Claude is a multi-agent autonomous coding framework that builds software th
 autonomous-coding/
 ├── apps/
 │   ├── backend/           # Python backend/CLI - ALL agent logic lives here
-│   │   ├── core/          # Client, auth, security
+│   │   ├── core/          # Client, auth, security, git provider detection
 │   │   ├── agents/        # Agent implementations
 │   │   ├── spec_agents/   # Spec creation agents
-│   │   ├── integrations/  # Graphiti, Linear, GitHub
+│   │   ├── runners/       # GitHub & GitLab automation
+│   │   ├── integrations/  # Graphiti, Linear
 │   │   └── prompts/       # Agent system prompts
 │   └── frontend/          # Electron desktop UI
 ├── guides/                # Documentation
@@ -195,6 +196,8 @@ See [RELEASE.md](RELEASE.md) for detailed release process documentation.
 **Integrations:**
 - **linear_updater.py** - Optional Linear integration for progress tracking
 - **runners/github/** - GitHub Issues & PRs automation
+- **runners/gitlab/** - GitLab Issues & MRs automation
+- **core/git_provider.py** - Multi-provider support (GitHub, GitLab) with auto-detection
 - **Electron MCP** - E2E testing integration for QA agents (Chrome DevTools Protocol)
   - Enabled with `ELECTRON_MCP_ENABLED=true` in `.env`
   - Allows QA agents to interact with running Electron app
@@ -237,7 +240,7 @@ main (user's branch)
 **Key principles:**
 - ONE branch per spec (`auto-claude/{spec-name}`)
 - Parallel work uses subagents (agent decides when to spawn)
-- NO automatic pushes to GitHub - user controls when to push
+- NO automatic pushes to remote (GitHub/GitLab) - user controls when to push
 - User reviews in spec worktree (`.worktrees/{spec-name}/`)
 - Final merge: spec branch → main (after user approval)
 
@@ -248,6 +251,65 @@ main (user's branch)
 4. User runs `--merge` to add to their project
 5. User pushes to remote when ready
 
+### Git Provider Support
+
+Auto Claude supports both **GitHub** and **GitLab** for pull request / merge request creation:
+
+**Configuration:**
+- **Per-project setting** in project settings UI or `.env` file
+- **Options**: Auto-detect (default), GitHub, GitLab
+- **Auto-detection**: Parses git remote URL to determine provider
+  - `github.com` → Uses GitHub (`gh pr create`)
+  - `gitlab.com` → Uses GitLab (`glab mr create`)
+  - Self-hosted GitLab instances also supported (e.g., `gitlab.mycompany.com`)
+  - Unknown/no remote → Defaults to GitHub
+
+**CLI Requirements:**
+- **GitHub**: Requires `gh` CLI installed and authenticated
+  - Install: `brew install gh` (macOS), `scoop install gh` (Windows), `sudo apt install gh` (Linux)
+  - Authenticate: `gh auth login`
+  - More info: https://cli.github.com/
+- **GitLab**: Requires `glab` CLI installed and authenticated
+  - Install: `brew install glab` (macOS), `scoop install glab` (Windows)
+  - Download: https://gitlab.com/gitlab-org/cli/-/releases
+  - Authenticate: `glab auth login`
+
+**Configuration Options:**
+
+*In Project Settings UI:*
+1. Open project settings
+2. Expand "Git Provider Settings"
+3. Choose: Auto-detect (recommended), GitHub, or GitLab
+
+*In `.env` file:*
+
+```bash
+# Set git provider (optional, defaults to auto-detect)
+GIT_PROVIDER=auto    # auto | github | gitlab
+```
+
+**Example Usage:**
+
+```bash
+# Auto-detect provider from git remote (default)
+python run.py --spec 001 --create-pr
+
+# Works automatically with both:
+# - [email protected]:user/repo.git → Creates GitHub PR using gh CLI
+# - [email protected]:user/repo.git → Creates GitLab MR using glab CLI
+```
+
+**Provider Detection Logic:**
+1. **Explicit setting** (if gitProvider is "github" or "gitlab") → Use that provider
+2. **Auto-detect** (if gitProvider is "auto" or not set) → Parse git remote URL
+3. **Fallback** (if no remote or unknown) → Default to GitHub
+
+**Implementation Details:**
+- Detection: `apps/backend/core/git_provider.py`
+- GitHub PR creation: `apps/backend/core/worktree.py` (`_create_github_pr()`)
+- GitLab MR creation: `apps/backend/core/worktree.py` (`_create_gitlab_mr()`)
+- UI settings: `apps/frontend/src/renderer/components/project-settings/IntegrationSettings.tsx`
+
 ### Contributing to Upstream
 
 **CRITICAL: When submitting PRs to AndyMik90/Auto-Claude, always target the `develop` branch, NOT `main`.**

PR_DESCRIPTION.md

@@ -0,0 +1,60 @@
+## Summary
+
+This PR implements comprehensive Conda environment management for Auto Claude, enabling isolated Python environments at both application and project levels. It also includes significant Windows/PowerShell compatibility fixes to ensure reliable terminal integration across platforms.
+
+### Key Features
+
+- **Conda Detection Service**: Automatically detects conda installations across OS-specific locations (miniconda, anaconda, mambaforge)
+- **Application-Level Environment**: Managed conda env at `~/miniconda3/envs/auto-claude` using `apps/backend/requirements.txt`
+- **Project-Level Environments**: Self-contained environments in `.envs/<project-name>/` within each project
+- **Automatic Terminal Activation**: Conda environments auto-activate when opening terminals for configured projects
+- **Cross-Platform Activation Scripts**: Generated scripts for CMD, PowerShell, and Bash
+- **VS Code Integration**: Auto-generated `.code-workspace` files with conda terminal profiles
+- **Python Version Detection**: Parses `requirements.txt`, `pyproject.toml`, and `environment.yml` for version constraints
+
+### Windows Compatibility Fixes
+
+- Added `windowsHide: true` to all spawn calls to prevent console window popups
+- Fixed PowerShell command syntax (`;` instead of `&&`, `$env:PATH=` instead of `PATH=`)
+- Platform-aware shell escaping (PowerShell uses `''` for quotes, bash uses `'\''`)
+- Added `pathWasModified` optimization to skip unnecessary PATH modifications (eliminates massive PATH echo)
+- PowerShell-specific conda activation using init scripts
+
+## Test plan
+
+- [ ] **Conda Detection**: Open App Settings > Paths, verify conda installation detected
+- [ ] **App Env Setup**: Click "Setup Auto Claude Environment", verify stepper completes
+- [ ] **Project Toggle**: Enable conda env for a project in Project Settings > General
+- [ ] **Project Env Setup**: Navigate to Python Env section, run setup, verify environment created
+- [ ] **Terminal Activation**: Open terminal for conda-enabled project, verify env activates automatically
+- [ ] **Windows Popup**: Verify no external PowerShell windows appear during operations
+- [ ] **Connect Claude (Windows)**: Click "Connect Claude", verify PowerShell syntax works without errors
+- [ ] **Settings Persistence**: Navigate away and back, verify all conda settings retained
+
+## Files Changed
+
+### New Files (11)
+- `conda-detector.ts` - Conda installation detection service
+- `conda-env-manager.ts` - Environment creation and management
+- `conda-workspace-generator.ts` - VS Code workspace file generation
+- `conda-project-structure.ts` - Project structure detection (pure-python vs mixed)
+- `conda-handlers.ts` - IPC handlers for conda operations
+- `conda-api.ts` - Preload API for renderer access
+- `PythonEnvSettings.tsx` - Project-level Python environment settings UI
+- `CondaSetupWizard.tsx` - Multi-step setup wizard with progress
+- `CondaDetectionDisplay.tsx` - Conda detection status display
+- `useCondaSetup.ts` - React hook for setup progress
+- `conda.ts` - TypeScript type definitions
+
+### Modified Files (48)
+- Terminal integration: `pty-manager.ts`, `terminal-lifecycle.ts`, `terminal-manager.ts`
+- Claude integration: `claude-cli-utils.ts`, `claude-integration-handler.ts`
+- Shell utilities: `shell-escape.ts`, `env-utils.ts`
+- Settings UI: `GeneralSettings.tsx`, `AppSettings.tsx`, `ProjectSettingsContent.tsx`
+- IPC handlers: Various handlers with `windowsHide` fixes
+- Type definitions: `settings.ts`, `project.ts`, `terminal.ts`, `ipc.ts`
+- i18n: English and French translation files
+
+---
+
+Generated with [Claude Code](https://claude.ai/code)

apps/backend/agents/coder.py

@@ -40,6 +40,7 @@
 from recovery import RecoveryManager
 from security.constants import PROJECT_DIR_ENV_VAR
 from task_logger import (
+    LogEntryType,
     LogPhase,
     get_task_logger,
 )
@@ -55,6 +56,7 @@
     print_key_value,
     print_status,
 )
+from user_feedback import get_unread_feedback
 
 from .base import AUTO_CONTINUE_DELAY_SECONDS, HUMAN_INTERVENTION_FILE
 from .memory_manager import debug_memory_system_status, get_graphiti_context
@@ -247,6 +249,30 @@
         subtask_id = next_subtask.get("id") if next_subtask else None
         phase_name = next_subtask.get("phase_name") if next_subtask else None
 
+        # Check for unread user feedback before starting this subtask
+        unread_feedback = get_unread_feedback(spec_dir)
+        feedback_text = None
+        if unread_feedback:
+            # Combine all unread feedback messages with their ACTUAL indices from full list
+            # unread_feedback is now a list of (actual_index, feedback_dict) tuples
+            feedback_messages = []
+            for actual_idx, fb in unread_feedback:
+                msg = fb.get("message", "")
+                timestamp = fb.get("timestamp", "")
+                # Use actual_idx so agent knows the correct index for mark_feedback_read tool
+                feedback_messages.append(f"[Index {actual_idx}] [{timestamp}]\n{msg}")
+            feedback_text = "\n\n".join(feedback_messages)
+
+            # Log feedback detection to task_logs.json (visible in UI)
+            task_logger = get_task_logger(spec_dir)
+            if task_logger:
+                task_logger.log(
+                    content=f"USER FEEDBACK DETECTED - {len(unread_feedback)} unread feedback item(s) will be incorporated into this subtask",
+                    entry_type=LogEntryType.INFO,
+                    phase=LogPhase.CODING,
+                    print_to_console=True,
+                )
+
         # Update status for this session
         status_manager.update_session(iteration)
         if phase_name:
@@ -476,6 +502,17 @@
             if sync_spec_to_source(spec_dir, source_spec_dir):
                 print_status("Implementation plan synced to main project", "success")
 
+            # Show plan statistics
+            from implementation_plan import ImplementationPlan
+
+            plan = ImplementationPlan.load(spec_dir / "implementation_plan.json")
+            total_phases = len(plan.phases)
+            total_subtasks = sum(len(p.subtasks) for p in plan.phases)
+            print_status(
+                f"Planning complete: {total_phases} phase(s), {total_subtasks} subtask(s)",
+                "success",
+            )
+
         # Handle session status
         if status == "complete":
             # Don't emit COMPLETE here - subtasks are done but QA hasn't run yet