Replace entire codebase with wish-ng implementation

This commit replaces all code with the wish-ng architecture, which includes:
- Consolidated package structure under packages/ directory
- New AI integration (wish-ai)
- Command & Control functionality (wish-c2 with Sliver)
- CLI interface (wish-cli)
- Core functionality (wish-core)
- Knowledge management (wish-knowledge)
- Comprehensive e2e testing framework
- Improved project structure and dependencies

Note: .claude/settings.local.json is not included (properly ignored)

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <noreply@anthropic.com>

Author: laysakura

.claude/commands/brc.md

@@ -0,0 +1,117 @@
+# +brc - Branch Review & Cleanup
+
+Review differences between current branch and main branch, and perform branch cleanup.
+
+## Overview
+
+This command performs comprehensive quality checks and cleanup before PR creation. It focuses on:
+
+- Ensuring code quality (lint, format)
+- Removing unnecessary files and code
+- **Complete cleanup of tmp/ directory**
+- Reviewing and resolving TODOs
+- Ensuring test success
+
+## Execution Steps
+
+### 1. Review Differences
+```bash
+git diff main...HEAD
+git diff --name-only main...HEAD
+```
+Review files and content changed in the current branch.
+
+### 2. Remove Unnecessary Files and Code Fragments
+- Remove debug print statements, console.log, temporary commented-out code
+- Remove temporary test files (test_*.py, *.log, *.tmp, etc.) not used in actual tests
+- Remove unused import statements
+- Remove experimental code only used during implementation
+- **tmp/ directory cleanup**:
+  ```bash
+  # Check tmp/ directory contents
+  if [ -d "tmp/" ]; then
+    echo "=== tmp/ directory contents ==="
+    ls -la tmp/
+    echo "Delete these files? (y/N)"
+    # Wait for user confirmation
+    read -r response
+    if [[ "$response" =~ ^[Yy]$ ]]; then
+      rm -rf tmp/*
+      echo "Cleaned up tmp/ directory"
+    fi
+  else
+    echo "tmp/ directory does not exist"
+  fi
+  ```
+
+### 3. Review and Resolve TODOs
+```bash
+# Search for TODO comments
+rg "TODO|FIXME|XXX|HACK" --type-add 'code:*.{py,ts,js,tsx,jsx}' -t code
+```
+- Review each TODO found
+- Immediately resolve items that should be addressed in current branch
+- For items to remain as future tasks, clearly document reasons in summary
+
+### 4. Fix Formatting
+```bash
+make format
+make lint
+```
+- If lint errors exist, review error messages and fix
+- Manually fix errors that cannot be auto-fixed
+
+### 5. Fix Tests
+```bash
+make test
+```
+- If tests fail:
+  - First analyze error messages
+  - Verify test expectations are correct
+  - Determine whether to fix implementation or tests
+  - If implementation should take priority over tests, document reasons in summary
+
+### 6. Final Review and Summary Display
+Display summary in the following format:
+
+```
+## Branch Review Summary
+
+### Changes
+- Changed files: X files
+- Added lines: +XXX
+- Deleted lines: -XXX
+
+### Cleanup Performed
+- [ ] Removed unnecessary debug code
+- [ ] Removed unused imports
+- [ ] Removed temporary files
+- [ ] Cleaned up tmp/ directory
+
+### TODO Status
+- Resolved: X items
+- Remaining: X items
+  - [Reason] TODO content
+
+### Test & Quality Check
+- [ ] make format: PASS/FAIL
+- [ ] make lint: PASS/FAIL  
+- [ ] make test: PASS/FAIL
+
+### Remaining Issues
+(Document if any)
+
+### Next Steps
+(If all checks PASS)
+Ready to create PR. Create PR with the following command:
+gh pr create --title "Title" --body "Description"
+```
+
+## Notes
+- Carefully judge unnecessary code to not compromise implementation intent
+- If tests fail, don't hastily fix tests; prioritize verifying implementation correctness
+- Don't automatically create PR; prompt for user's final confirmation
+- **About tmp/ directory**:
+  - All intermediate files created by Claude Code (test files, planning documents, etc.) are placed in tmp/
+  - Always perform cleanup when executing +brc
+  - Request user confirmation before deletion

.claude/commands/e2e.md

@@ -0,0 +1,101 @@
+# +e2e: End-to-End Test Execution
+
+This shortcut executes phased end-to-end tests for wish-cli-basic-ui.
+
+## Execution Command
+
+```bash
+make e2e
+```
+
+## Usage
+
+```bash
++e2e    # wish CLI e2e test (all phases)
+```
+
+## Test Configuration
+
+### Level 1: Component Integration Tests
+- ✅ Basic application startup/shutdown tests
+- ✅ `/help` command functionality verification
+- ✅ `/status` command functionality verification
+- ✅ `/mode` command functionality verification
+
+### Level 2: Workflow Integration Tests
+- ✅ Invalid command error handling
+- ✅ Invalid mode error handling
+- ✅ Multiple command sequential execution
+- ✅ Session state consistency verification
+- ✅ Recovery functionality after errors
+
+## Execution Methods
+
+**Execute from Claude Code:**
+```bash
++e2e
+```
+
+**Phased execution:**
+```bash
+make e2e-component    # Level 1 only
+make e2e-workflow     # Level 2 only
+make e2e             # All phases
+```
+
+**Direct pytest execution:**
+```bash
+uv run pytest e2e-tests/component/ -v
+uv run pytest e2e-tests/workflows/ -v
+```
+
+## Test Result Interpretation
+
+### Success Output Example
+```
+🚀 Starting wish-cli-basic-ui E2E Tests
+==================================================
+🧪 Running: test_basic_startup
+✅ PASS basic_startup (2.31s): Application started and exited successfully
+🧪 Running: test_help_command
+✅ PASS help_command (1.87s): Help command displayed expected content
+...
+✅ ALL TESTS PASSED
+```
+
+### Handling Failures
+
+**Common Issues:**
+1. **Startup failure**: Reinstall dependencies with `uv sync --dev`
+2. **Timeout**: Check system load
+3. **Command errors**: Verify latest code is deployed
+
+**Debug Methods:**
+```bash
+# Run tests with detailed output
+python scripts/e2e_test.py --verbose
+
+# Manual functionality verification
+uv run wish
+```
+
+## Continuous Quality Management
+
+This E2E test is executed in the following situations:
+- Functionality verification after adding new features
+- Regression testing after bug fixes
+- Quality verification before PR creation
+- Automated execution in CI/CD pipeline
+
+## Expected Behavior
+
+- **All tests pass**: Basic functionality working normally
+- **Partial failures**: Related functionality needs fixing
+- **All tests fail**: Possible basic environment issues
+
+## Notes
+
+- Test execution time: Approximately 10-30 seconds
+- Minimal external dependencies (uses mocks)
+- No actual network connection required
+- Environment cleanup is automatic

.claude/commands/fixci.md

@@ -0,0 +1,20 @@
+Investigate and fix CI failures for the Pull Request associated with the current branch.
+
+Follow these steps:
+
+1.  **Identify PR information**: Execute `gh pr list --head $(git branch --show-current)` to identify the PR number and URL associated with the current branch. If no PR is found, prompt the user to create a PR.
+
+2.  **Check CI status**: Use `gh pr checks` to display a summary of CI status for the identified PR.
+
+3.  **Analyze failure logs**: Combine `gh run list` and `gh run view --log-failed` to retrieve and display error logs from failed jobs in the latest workflow run.
+
+4.  **Root cause analysis and fix proposal**:
+    - Analyze the displayed error logs to identify the root cause of failures (e.g., test failures, lint errors, build issues).
+    - Based on the identified cause, develop a specific fix strategy.
+    - Present the fix strategy to the user and obtain approval before proceeding with code modifications.
+
+5.  **Fix and verify**:
+    - Modify code according to the approved strategy.
+    - After fixes, run `make lint` and `make test` locally to verify the issues are resolved.
+
+6.  **Commit and push**: Commit the fixes and push to remote to trigger CI re-run.

.claude/commands/fixlint.md

@@ -0,0 +1,53 @@
+# +fixlint: Lint Fix Shortcut
+
+This shortcut executes `make format` and automatically attempts to fix lint errors if they occur.
+
+## Execution Command
+
+```bash
+make format
+```
+
+## Purpose
+
+- Execute code formatting for all workspaces
+- Automatically fix lint errors
+- Ensure compliance with code quality standards
+
+## Usage
+
+```bash
++fixlint
+```
+
+## Fix Process
+
+1. **Initial format execution**: Format all workspace code with `make format`
+2. **Lint check**: Check for lint errors with `make lint`
+3. **Auto-fix**: Automatically fix the following types of lint errors:
+   - Import order errors
+   - Unused imports
+   - Line length violations
+   - Indentation errors
+   - Whitespace errors
+   - Naming convention violations
+   - Code style violations
+   - Docstring format errors
+4. **Re-run**: Execute lint check again after fixes
+5. **Iterate**: Repeat steps 2-4 until all lint errors are resolved
+
+## Target
+
+- Entire wish source code
+
+## Tools Used
+
+- **ruff**: Fast Python linter/formatter
+- **Auto-fix feature**: Automatic fixes via `ruff check --fix`
+
+## Notes
+
+- Formatting changes do not affect existing logic
+- Some complex lint errors may require manual fixes
+- Recommended to run before committing
+- For large changes, review diffs before committing

.claude/commands/fixtest.md

@@ -0,0 +1,46 @@
+# +fixtest: Test Fix Shortcut
+
+This shortcut executes `make test` and automatically attempts to fix test errors if they occur.
+
+## Execution Command
+
+```bash
+make test
+```
+
+## Purpose
+
+- Execute unit tests for all workspaces
+- Automatically fix test errors
+- Iterate until all tests pass
+
+## Usage
+
+```bash
++fixtest
+```
+
+## Fix Process
+
+1. **Initial test execution**: Run tests for all workspaces with `make test`
+2. **Error analysis**: Identify the cause of failed tests
+3. **Auto-fix**: Automatically fix the following types of errors:
+   - Import errors
+   - Errors due to module name changes
+   - Errors due to package structure changes
+   - Type errors
+   - Syntax errors
+   - Dependency errors
+4. **Re-run**: Execute tests again after fixes
+5. **Iterate**: Repeat steps 2-4 until all tests pass
+
+## Target
+
+- Entire wish test suite
+
+## Notes
+
+- LLM tests (`@pytest.mark.llm`) are excluded
+- Time-consuming tests (`@pytest.mark.slow`) are excluded
+- Test fixes are executed carefully to avoid breaking existing logic
+- Complex business logic errors may require manual review

.claude/commands/pr.md

@@ -0,0 +1,20 @@
+Create a Pull Request from the current branch to the `main` branch following best practices.
+If `draft` is specified in `$ARGUMENTS`, create as a draft PR.
+
+Follow these steps:
+
+1.  **Pre-checks**:
+    - Use `git status` to check for uncommitted changes. If any exist, ask the user whether to commit or stash them.
+    - Use `git branch --show-current` to verify the current branch name is not `main` or `master`.
+
+2.  **Sync with remote**:
+    - Execute `git fetch origin`.
+    - If the local branch doesn't exist on remote or has unpushed commits, execute `git push -u origin $(git branch --show-current)` to sync with remote.
+
+3.  **Generate PR information**:
+    - **Title**: If there's only one commit in the diff with `main` branch, use the first line of that commit message as the title. If there are multiple commits, convert the branch name to a human-readable format (e.g., `feat/new-feature` -> `Feat: New Feature`) for the title.
+    - **Body**: Include the list of commits in the diff with `main` branch in bullet point format. Also provide template sections for "Related Issues" and "Testing".
+
+4.  **Create PR**:
+    - Use the `gh pr create` command to create the PR.
+    - If `$ARGUMENTS` contains `draft`, add the `--draft` flag.

.claude/settings.json

@@ -0,0 +1,28 @@
+{
+  "hooks": {
+    "PreToolUse": [
+      {
+        "matcher": "Bash",
+        "hooks": [
+          {
+            "type": "command",
+            "command": "~/.local/bin/enforce-uv-claude-hook"
+          }
+        ]
+      },
+      {
+        "matcher": "Edit|Write|MultiEdit",
+        "hooks": [
+          {
+            "type": "command",
+            "command": "~/.local/bin/enforce-no-serialization-claude-hook"
+          },
+          {
+            "type": "command",
+            "command": "~/.local/bin/restrict-working-directory-claude-hook"
+          }
+        ]
+      }
+    ]
+  }
+}