v2.7.0-alpha.1 - ReasoningBank Integration Fix
v2.7.0-alpha.1 - ReasoningBank Integration Fix
π Critical Bug Fix
Fixed: ReasoningBank store and query commands now work correctly
What Was Broken (v2.7.0-alpha)
# This failed with "--agent required" error
npx claude-flow@alpha memory store api_pattern "Use env vars" --reasoningbank
β Error: --agent required
What's Fixed (v2.7.0-alpha.1)
# Now works correctly
npx claude-flow@alpha memory store api_pattern "Use env vars" --reasoningbank
β
Stored successfully in ReasoningBank
π§ Memory ID: mem_...
π Semantic search: enabled
π§ Technical Changes
- Direct SDK Integration: Replaced broken CLI delegation with direct agentic-flow SDK calls
- Dynamic Model Support: Embedding model now respects ReasoningBank configuration
- Proper Error Handling: Graceful fallback when embeddings fail
- Database Schema Compliance: All required fields (model, dims) now included
β οΈ Performance Characteristics
Important: Reasoning Bank embedding computation can be slow (10-45 seconds per operation)
This is a known limitation of the semantic search feature:
- Memory storage: Always succeeds immediately
- Embedding computation: Runs asynchronously, may be slow
- If embeddings fail: Memory is still stored, query fallback works
Recommendations
Use ReasoningBank for:
- High-value knowledge (API patterns, best practices)
- Memories you'll search semantically
- Long-term learning storage
Use Basic Mode for:
- Frequent, quick operations
- Simple key-value storage
- Performance-critical workflows
# Fast: Basic mode (< 1ms)
claude-flow memory store quick_note "Build passed"
# Slower: ReasoningBank with semantic search (10-45s)
claude-flow memory store api_pattern "Use env vars" --reasoningbank
β What Works
- β
memory init --reasoningbank- Initialize database - β
memory store key "value" --reasoningbank- Store memories - β
memory query "search" --reasoningbank- Query with semantic search (or fallback) - β
memory status --reasoningbank- Show statistics - β
memory list --reasoningbank- List all memories - β Basic mode (default) - Still fast and reliable
π Upgrade from v2.7.0-alpha
# Clear npm cache to avoid ENOTEMPTY error
rm -rf ~/.npm/_npx
# Install new version
npm install -g claude-flow@alpha
# Or use npx
npx claude-flow@alpha --version
# Should show: v2.7.0-alpha.1
π Validation Results
- β Init: Works
- β Store: Works (memory stored immediately, embeddings async)
- β Status: Works
- β Query: Works (with fallback for missing embeddings)
- β Basic Mode: Works (backward compatible)
- β οΈ Embeddings: Slow but functional
π― Quick Start
# 1. Initialize ReasoningBank (one-time)
claude-flow memory init --reasoningbank
# 2. Store important knowledge
claude-flow memory store api_security "Always validate input, use prepared statements" --reasoningbank
# 3. Check status
claude-flow memory status --reasoningbank
# 4. For fast operations, use basic mode (default)
claude-flow memory store build_log "Tests passed"
π Known Issues
-
Embedding Performance: 10-45 seconds per operation
- Impact: Slower than basic mode
- Workaround: Use basic mode for frequent operations
- Status: This is an agentic-flow/ReasoningBank limitation
-
npx Cache Error: ENOTEMPTY on upgrade
- Impact: npx installation may fail
- Workaround:
rm -rf ~/.npm/_npxor use global install - Status: npm bug, not claude-flow specific
π Summary
Status: β Production Ready with Performance Notes
This release fixes the critical ReasoningBank integration bug. The feature now works correctly, though users should be aware of performance characteristics and choose the appropriate mode for their use case.
Recommendation: Use ReasoningBank selectively for high-value knowledge where semantic search justifies the latency. Use basic mode for fast, frequent operations.