# Real Project Testing Prompts
## Prompt 1: Commission Processing Vendor Extractors
```
Test config-based project tracking with the commission-processing-vendor-extractors project. Execute these commands:
1. Navigate to the codebase-mcp project:
cd /Users/cliffclarke/Claude_Code/codebase-mcp
2. Create config file for commission-processing project:
cat > /Users/cliffclarke/Claude_Code/commission-processing-vendor-extractors/.codebase-mcp/config.json <<'EOF'
{
"version": "1.0",
"project": {
"name": "commission-processing-vendor-extractors",
"id": "commission-vendor-extractors"
},
"auto_switch": true,
"description": "Commission processing vendor file extraction system"
}
EOF
3. Verify the config file was created and is valid:
uv run python -c "
from src.auto_switch.validation import validate_config_syntax
from pathlib import Path
config = validate_config_syntax(Path('/Users/cliffclarke/Claude_Code/commission-processing-vendor-extractors/.codebase-mcp/config.json'))
print(f'✓ Config valid')
print(f' Project name: {config[\"project\"][\"name\"]}')
print(f' Project ID: {config[\"project\"][\"id\"]}')
"
4. Test the complete workflow - set working directory and resolve project:
uv run python -c "
import asyncio
from unittest.mock import Mock
from src.database.session import resolve_project_id
from src.auto_switch.session_context import get_session_context_manager
from src.mcp.tools.project import set_working_directory
async def test():
# Start session manager
mgr = get_session_context_manager()
await mgr.start()
# Create mock context for session
ctx = Mock()
ctx.session_id = 'commission-processing-session'
# Call set_working_directory tool
result = await set_working_directory.fn(
directory='/Users/cliffclarke/Claude_Code/commission-processing-vendor-extractors',
ctx=ctx
)
print(f'✓ set_working_directory tool response:')
print(f' Session ID: {result[\"session_id\"]}')
print(f' Config found: {result[\"config_found\"]}')
print(f' Project name: {result[\"project_info\"][\"name\"]}')
print(f' Project ID: {result[\"project_info\"][\"id\"]}')
# Test that resolve_project_id uses the config
resolved = await resolve_project_id(explicit_id=None, ctx=ctx)
print(f'\\n✓ Project resolution:')
print(f' Resolved to: {resolved}')
assert resolved == 'commission-vendor-extractors'
await mgr.stop()
print('\\n✓ Commission processing project configured successfully!')
asyncio.run(test())
"
5. Now test indexing the repository (this will actually index the codebase):
uv run python -c "
import asyncio
from unittest.mock import Mock
from src.mcp.tools.indexing import index_repository
from src.auto_switch.session_context import get_session_context_manager
async def test_indexing():
mgr = get_session_context_manager()
await mgr.start()
ctx = Mock()
ctx.session_id = 'commission-processing-session'
ctx.info = Mock(return_value=asyncio.coroutine(lambda x: None)())
# Set working directory first
await mgr.set_working_directory(
ctx.session_id,
'/Users/cliffclarke/Claude_Code/commission-processing-vendor-extractors'
)
print('Indexing commission-processing-vendor-extractors...')
print('(This may take 30-60 seconds depending on codebase size)')
# Index WITHOUT explicit project_id - should use config automatically
result = await index_repository.fn(
repo_path='/Users/cliffclarke/Claude_Code/commission-processing-vendor-extractors',
project_id=None, # Let it resolve from config
force_reindex=False,
ctx=ctx
)
print(f'\\n✓ Indexing complete:')
print(f' Repository ID: {result[\"repository_id\"]}')
print(f' Files indexed: {result[\"files_indexed\"]}')
print(f' Chunks created: {result[\"chunks_created\"]}')
print(f' Duration: {result[\"duration_seconds\"]:.2f}s')
print(f' Project ID: {result[\"project_id\"]}')
print(f' Schema: {result[\"schema_name\"]}')
assert result['project_id'] == 'commission-vendor-extractors'
assert result['schema_name'] == 'project_commission_vendor_extractors'
await mgr.stop()
print('\\n✓ Commission processing codebase indexed with auto-switch!')
asyncio.run(test_indexing())
"
Report the results. The indexing should use project_id "commission-vendor-extractors" without explicitly passing it.
```
---
## Prompt 2: Workflow MCP Project
```
Test config-based project tracking with the workflow-mcp project. This tests multi-project isolation. Execute these commands:
1. Navigate to the codebase-mcp project:
cd /Users/cliffclarke/Claude_Code/codebase-mcp
2. Create config file for workflow-mcp project:
cat > /Users/cliffclarke/Claude_Code/workflow-mcp/.codebase-mcp/config.json <<'EOF'
{
"version": "1.0",
"project": {
"name": "workflow-mcp",
"id": "workflow-mcp-server"
},
"auto_switch": true,
"description": "Workflow MCP server for project tracking and work item management"
}
EOF
3. Verify the config file was created and is valid:
uv run python -c "
from src.auto_switch.validation import validate_config_syntax
from pathlib import Path
config = validate_config_syntax(Path('/Users/cliffclarke/Claude_Code/workflow-mcp/.codebase-mcp/config.json'))
print(f'✓ Config valid')
print(f' Project name: {config[\"project\"][\"name\"]}')
print(f' Project ID: {config[\"project\"][\"id\"]}')
"
4. Test multi-session isolation with both projects:
uv run python -c "
import asyncio
from unittest.mock import Mock
from src.database.session import resolve_project_id
from src.auto_switch.session_context import get_session_context_manager
async def test():
mgr = get_session_context_manager()
await mgr.start()
# Session 1: Commission processing
ctx1 = Mock()
ctx1.session_id = 'commission-session'
await mgr.set_working_directory(
'commission-session',
'/Users/cliffclarke/Claude_Code/commission-processing-vendor-extractors'
)
# Session 2: Workflow MCP
ctx2 = Mock()
ctx2.session_id = 'workflow-session'
await mgr.set_working_directory(
'workflow-session',
'/Users/cliffclarke/Claude_Code/workflow-mcp'
)
# Resolve both sessions
result1 = await resolve_project_id(explicit_id=None, ctx=ctx1)
result2 = await resolve_project_id(explicit_id=None, ctx=ctx2)
print(f'✓ Multi-session isolation test:')
print(f' Session 1 resolved to: {result1}')
print(f' Session 2 resolved to: {result2}')
# Verify they're different
assert result1 == 'commission-vendor-extractors', f'Expected commission-vendor-extractors, got {result1}'
assert result2 == 'workflow-mcp-server', f'Expected workflow-mcp-server, got {result2}'
# Verify session 1 still resolves correctly after session 2 was added
result1_check = await resolve_project_id(explicit_id=None, ctx=ctx1)
assert result1_check == 'commission-vendor-extractors'
print(f'\\n✓ Sessions are completely isolated!')
await mgr.stop()
asyncio.run(test())
"
5. Index the workflow-mcp repository:
uv run python -c "
import asyncio
from unittest.mock import Mock
from src.mcp.tools.indexing import index_repository
from src.auto_switch.session_context import get_session_context_manager
async def test_indexing():
mgr = get_session_context_manager()
await mgr.start()
ctx = Mock()
ctx.session_id = 'workflow-session'
ctx.info = Mock(return_value=asyncio.coroutine(lambda x: None)())
# Set working directory for workflow-mcp
await mgr.set_working_directory(
ctx.session_id,
'/Users/cliffclarke/Claude_Code/workflow-mcp'
)
print('Indexing workflow-mcp...')
print('(This may take 30-60 seconds depending on codebase size)')
# Index WITHOUT explicit project_id - should use config automatically
result = await index_repository.fn(
repo_path='/Users/cliffclarke/Claude_Code/workflow-mcp',
project_id=None, # Let it resolve from config
force_reindex=False,
ctx=ctx
)
print(f'\\n✓ Indexing complete:')
print(f' Repository ID: {result[\"repository_id\"]}')
print(f' Files indexed: {result[\"files_indexed\"]}')
print(f' Chunks created: {result[\"chunks_created\"]}')
print(f' Duration: {result[\"duration_seconds\"]:.2f}s')
print(f' Project ID: {result[\"project_id\"]}')
print(f' Schema: {result[\"schema_name\"]}')
assert result['project_id'] == 'workflow-mcp-server'
assert result['schema_name'] == 'project_workflow_mcp_server'
await mgr.stop()
print('\\n✓ Workflow-mcp codebase indexed with auto-switch!')
asyncio.run(test_indexing())
"
6. Verify both projects are indexed in separate schemas:
uv run python -c "
import asyncio
from src.database.session import get_session
async def verify():
# Check commission-processing schema
async with get_session(project_id='commission-vendor-extractors') as db:
result = await db.execute('SELECT COUNT(*) FROM repositories')
count1 = result.scalar()
print(f'Commission processing schema has {count1} repository')
# Check workflow-mcp schema
async with get_session(project_id='workflow-mcp-server') as db:
result = await db.execute('SELECT COUNT(*) FROM repositories')
count2 = result.scalar()
print(f'Workflow MCP schema has {count2} repository')
print(f'\\n✓ Both projects indexed in separate schemas!')
asyncio.run(verify())
"
Report the results. Both projects should be indexed in completely separate database schemas with full isolation.
```
---
## Expected Results
After running both prompts, you should see:
### Commission Processing Project:
- Config file created at `/Users/cliffclarke/Claude_Code/commission-processing-vendor-extractors/.codebase-mcp/config.json`
- Project ID resolves to: `commission-vendor-extractors`
- Schema name: `project_commission_vendor_extractors`
- Repository indexed with X files and Y chunks
### Workflow MCP Project:
- Config file created at `/Users/cliffclarke/Claude_Code/workflow-mcp/.codebase-mcp/config.json`
- Project ID resolves to: `workflow-mcp-server`
- Schema name: `project_workflow_mcp_server`
- Repository indexed with X files and Y chunks
### Session Isolation:
- Two concurrent sessions maintain separate project contexts
- No cross-contamination between projects
- Each session correctly resolves to its own project ID
### Performance:
- Config discovery: <50ms
- Indexing: 30-120 seconds depending on codebase size
- Session resolution: <1ms
Both codebases will be indexed in completely separate PostgreSQL schemas, demonstrating full multi-project isolation!
