Skip to main content
Glama

Codebase MCP Server

by Ravenight13
TEST_WORKFLOW_MCP_INDEXING.md6.67 kB
# Test Prompt: Index workflow-mcp Project with Background Indexing **Purpose**: Test background indexing with a real medium-sized codebase (workflow-mcp) **Branch**: `015-background-indexing-mvp` (on codebase-mcp) --- ## Copy This Prompt for workflow-mcp Chat ``` I want to test the background indexing feature from the codebase-mcp project by indexing the entire workflow-mcp codebase. **Setup**: - The codebase-mcp MCP server should be running with branch: 015-background-indexing-mvp - I want to index the workflow-mcp project located at: /Users/cliffclarke/Claude_Code/workflow-mcp - This will test background indexing with a real medium-sized Python project **Steps**: 1. **Start Background Indexing**: - Use the codebase-mcp tool: start_indexing_background() - Repository path: /Users/cliffclarke/Claude_Code/workflow-mcp - Project ID: "workflow-mcp-test" - Save the returned job_id 2. **Monitor Progress**: - Poll status every 5 seconds using: get_indexing_status(job_id) - Show me each status update with: - Current status (pending/running/completed/failed) - Files indexed so far - Chunks created so far - Elapsed time - Continue until status is "completed" or "failed" 3. **Report Results**: When complete, show me: - Total files indexed - Total chunks created - Total duration (from started_at to completed_at) - Any errors encountered - Average indexing speed (files per second) **Expected Results**: - workflow-mcp has approximately 50-100 Python files - Should complete in 1-3 minutes - Should create 500-2000 chunks depending on code complexity - No timeout errors (background indexing should handle the duration) **What I'm Testing**: ✅ Background indexing works with real medium-sized project ✅ No MCP timeout errors (job runs in background) ✅ Status polling provides accurate progress updates ✅ State transitions correctly (pending → running → completed) ✅ Metrics are accurate for real codebase ✅ Cross-project indexing works (codebase-mcp indexing workflow-mcp) Please execute this test and show me detailed progress updates every 5 seconds until completion. ``` --- ## Alternative: Minimal Prompt If you want a shorter version: ``` Use codebase-mcp's background indexing to index the workflow-mcp project: 1. Start: start_indexing_background(repo_path="/Users/cliffclarke/Claude_Code/workflow-mcp", project_id="workflow-mcp-test") 2. Poll every 5s: get_indexing_status(job_id) until status="completed" 3. Show: status updates, final metrics (files, chunks, duration) Expected: 50-100 files, 1-3 minutes, no timeout errors. ``` --- ## What to Look For ### Success Indicators ✅ - Job creates immediately (<1 second) - Status transitions: `pending` → `running` → `completed` - Files indexed: 50-100 (actual count from workflow-mcp) - Chunks created: 500-2,000 (depends on code size) - Duration: 1-3 minutes (no timeout!) - No errors in error_message field ### Potential Issues ⚠️ - **Job stuck in "pending"**: Worker may not have started - **Job fails quickly**: Path might be wrong, check repo_path - **Very slow progress**: Normal for first-time indexing with embeddings - **Timeout error**: Should NOT happen (that's what we're testing!) --- ## Expected Output Pattern ``` Starting background indexing... ✅ Job created: a7f3b2c1-4d5e-6f7a-8b9c-0d1e2f3a4b5c Polling status every 5 seconds... [0s] Status: pending Files: 0, Chunks: 0 [5s] Status: running Files: 12, Chunks: 156 Progress: ~12 files/5s = 2.4 files/sec [10s] Status: running Files: 28, Chunks: 374 Progress: ~16 files/5s = 3.2 files/sec [15s] Status: running Files: 45, Chunks: 628 Progress: ~17 files/5s = 3.4 files/sec [20s] Status: running Files: 63, Chunks: 891 Progress: ~18 files/5s = 3.6 files/sec [25s] Status: completed ✅ Files: 78, Chunks: 1,142 Duration: 25.3 seconds Final Results: ✅ Successfully indexed workflow-mcp codebase - Files indexed: 78 - Chunks created: 1,142 - Duration: 25.3s - Average speed: 3.1 files/second - Status: completed - Errors: None ``` --- ## Validation Checklist After the test completes, verify: - [ ] Job created in <1 second ✓ - [ ] No MCP client timeout (job ran in background) ✓ - [ ] Status transitioned correctly (pending → running → completed) ✓ - [ ] Files indexed matches workflow-mcp file count ✓ - [ ] Chunks created is reasonable (10-20 chunks per file) ✓ - [ ] Duration is reasonable (1-3 minutes for ~100 files) ✓ - [ ] No errors in error_message field ✓ - [ ] All timestamps populated (created_at, started_at, completed_at) ✓ - [ ] Can query the indexed data with semantic search ✓ --- ## Follow-Up Test (Optional) After indexing completes, test semantic search on the indexed workflow-mcp code: ``` Now that workflow-mcp is indexed, test semantic search: Search for: "database session management" Expected: Should find results from workflow-mcp code about database sessions Search for: "MCP tool registration" Expected: Should find workflow-mcp MCP tool implementation code This validates that: 1. Background indexing completed successfully 2. The indexed data is searchable 3. Semantic search works across the indexed codebase ``` --- ## Troubleshooting ### "Repository not found" error **Fix**: Verify the path is correct: ```bash ls -la /Users/cliffclarke/Claude_Code/workflow-mcp # Should show workflow-mcp project files ``` ### Job stays in "pending" forever **Fix**: Check codebase-mcp server logs for worker errors ### "Table indexing_jobs does not exist" **Fix**: Ensure codebase-mcp is on branch `015-background-indexing-mvp` and migration is applied: ```bash cd /Users/cliffclarke/Claude_Code/codebase-mcp git checkout 015-background-indexing-mvp alembic upgrade head ``` ### Very slow indexing (>5 minutes) **Possible causes**: - First-time embedding generation (normal) - Large files taking time to process - Database write contention --- ## Success Criteria For this test to be considered successful: 1. ✅ **No Timeout**: Job completes without MCP client timeout (proves background processing works) 2. ✅ **Accurate Metrics**: Files indexed matches actual workflow-mcp file count 3. ✅ **Reasonable Speed**: 1-5 files/second (accounting for embedding generation) 4. ✅ **State Management**: Clean transitions through pending → running → completed 5. ✅ **No Errors**: error_message field is null 6. ✅ **Searchable**: Can search the indexed workflow-mcp code afterward If all criteria met: **Background Indexing MVP validated with real-world project! 🎉**

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/Ravenight13/codebase-mcp'

If you have feedback or need assistance with the MCP directory API, please join our Discord server