agentic-flow 1.7.0 → 1.7.3

package/CHANGELOG.md

@@ -5,6 +5,123 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [1.7.3] - 2025-10-19
+
+### Fixed
+- **CRITICAL: Skills Installation Path** - Fixed skills to install at top level for Claude Code discovery
+  - **Before**: Skills installed to `.claude/skills/agentic-flow/[skill-name]/` (WRONG - nested subdirectory)
+  - **After**: Skills install to `.claude/skills/[skill-name]/` (CORRECT - top level)
+  - Claude Code requires skills at top level, NOT in subdirectories
+  - Affects `npx agentic-flow skills create` command
+  - All 4 example skills (agentdb-vector-search, agentdb-memory-patterns, swarm-orchestration, reasoningbank-intelligence) now install correctly
+
+### Migration
+If you installed skills with v1.7.2:
+```bash
+# Move skills from subdirectory to top level
+cd .claude/skills
+mv agentic-flow/* .
+rmdir agentic-flow
+# Restart Claude Code
+```
+
+Or simply reinstall:
+```bash
+rm -rf .claude/skills
+npx agentic-flow@latest skills create
+```
+
+## [1.7.2] - 2025-10-19
+
+### Added
+- **Skills System Integration** - Full CLI support for Claude Code Skills management
+  - `npx agentic-flow skills list` - List all installed skills (personal + project)
+  - `npx agentic-flow skills init` - Initialize skills directories (~/.claude/skills and .claude/skills)
+  - `npx agentic-flow skills create` - Create agentic-flow example skills (AgentDB, swarm, reasoningbank)
+  - `npx agentic-flow skills init-builder` - Install skill-builder framework for creating custom skills
+  - `npx agentic-flow skills help` - Show comprehensive skills command documentation
+  - Skills automatically discovered by Claude Code across all surfaces (Claude.ai, CLI, SDK, API)
+
+### Changed
+- **README Updates** - Added comprehensive Skills System documentation
+  - New Skills System section with Quick Start guide
+  - Updated Quick Navigation table to highlight Skills
+  - Updated CLI Commands section with all 5 skills commands
+  - Added skill structure examples with YAML frontmatter
+  - Listed 5 built-in skills: skill-builder, agentdb-memory-patterns, agentdb-vector-search, reasoningbank-intelligence, swarm-orchestration
+
+### Documentation
+- Updated both main README.md and npm package README.md with identical Skills documentation
+- Added skills to Quick Navigation for easy discovery
+- Included benefits: reusable, discoverable, structured, progressive, validated
+
+## [1.7.1] - 2025-10-19
+
+### Fixed
+- **CRITICAL: Skills Top-Level Installation** - Fixed skills to install at top level for Claude Code discovery
+  - **Claude Code requires**: `~/.claude/skills/[skill-name]/` (top level)
+  - **NOT supported**: `~/.claude/skills/namespace/[skill-name]/` (nested subdirectories)
+  - All skills now correctly install at top level: `~/.claude/skills/skill-builder/`, `~/.claude/skills/swarm-orchestration/`, etc.
+  - Skills are now properly discovered by Claude Code after restart
+  - Reverted incorrect v1.7.0 namespace changes that broke skill discovery
+
+### Changed
+- Updated `skills-manager.ts` to install all skills at top level (removed nested namespace)
+- Moved skill source files from `agentic-flow/` subdirectory to top level in `.claude/skills/`
+- Simplified source detection paths
+
+### Documentation
+- Added `docs/plans/skills/SKILLS_TOP_LEVEL_FIX.md` - Explanation of fix and testing
+- Updated `docs/plans/skills/SKILL_INSTALLATION_ANALYSIS.md` - Corrected with top-level requirement
+- Deprecated `docs/plans/skills/MIGRATION_v1.7.0.md` - Namespace approach was incorrect
+
+### Migration
+If you installed skills with v1.7.0:
+```bash
+# Move skills from namespace to top level
+cd ~/.claude/skills
+for skill in agentic-flow/*; do
+  mv "agentic-flow/$skill" "$(basename $skill)"
+done
+rm -rf agentic-flow
+# Restart Claude Code
+```
+
+Or simply reinstall:
+```bash
+npx agentic-flow skills init personal --with-builder
+```
+
+## [1.6.6] - 2025-10-18
+
+### Changed
+
+- **AgentDB v1.0.5** - Updated to latest version with browser bundle support
+  - Now includes `dist/agentdb.min.js` (196KB) and `dist/agentdb.js` (380KB) for CDN usage
+  - Browser bundles support direct import via unpkg/jsDelivr
+  - Source maps included for debugging
+  - WASM backend fully functional in browser environments
+
+## [1.6.5] - 2025-10-18
+
+### Changed
+
+- **AgentDB Dependency** - Updated to use published npm package instead of local file reference
+  - Changed from `"agentdb": "file:../packages/agentdb"` to `"agentdb": "^1.0.4"`
+  - Now uses stable published version from npm registry
+  - Includes all 20 MCP tools (10 core AgentDB + 10 learning tools)
+  - Easier installation and dependency management for end users
+
+### Added
+
+- **AgentDB v1.0.4 Integration** - Complete MCP learning system now available
+  - 10 learning tools: learning_start_session, learning_end_session, learning_predict, learning_feedback, learning_train, learning_metrics, learning_transfer, learning_explain, experience_record, reward_signal
+  - Q-learning with epsilon-greedy exploration
+  - Multi-dimensional reward system (success 40%, efficiency 30%, quality 20%, cost 10%)
+  - Experience replay buffer with prioritized sampling
+  - Transfer learning between similar tasks
+  - Session management with state persistence
+
 ## [1.6.4] - 2025-10-16
 
 ### 🚀 QUIC Transport - Production Ready (100% Complete)

package/README.md

@@ -17,7 +17,7 @@
 |-------------|---------------|---------------|
 | [Quick Start](#-quick-start) | [Agent Booster](#-core-components) | [Agent List](#-agent-types) |
 | [Deployment Options](#-deployment-options) | [ReasoningBank](#-core-components) | [MCP Tools](#-mcp-tools-213-total) |
-| [Model Optimization](#-model-optimization) | [Multi-Model Router](#-using-the-multi-model-router) | [Complete Docs](https://github.com/ruvnet/agentic-flow/tree/main/docs) |
+| [Model Optimization](#-model-optimization) | [Skills System](#-skills-system-claude-code-integration) | [Complete Docs](https://github.com/ruvnet/agentic-flow/tree/main/docs) |
 
 ---
 
@@ -53,33 +53,19 @@ Most AI coding agents are **painfully slow** and **frustratingly forgetful**. Th
 | Component | Description | Performance | Documentation |
 |-----------|-------------|-------------|---------------|
 | **Agent Booster** | Ultra-fast local code transformations via Rust/WASM (auto-detects edits) | 352x faster, $0 cost | [Docs](https://github.com/ruvnet/agentic-flow/tree/main/agent-booster) |
-| **AgentDB** | State-of-the-art memory with causal reasoning, reflexion, and skill learning | p95 < 50ms, 80% hit rate | [Docs](./src/agentdb/README.md) |
 | **ReasoningBank** | Persistent learning memory system with semantic search | 46% faster, 100% success | [Docs](https://github.com/ruvnet/agentic-flow/tree/main/agentic-flow/src/reasoningbank) |
 | **Multi-Model Router** | Intelligent cost optimization across 100+ LLMs | 85-99% cost savings | [Docs](https://github.com/ruvnet/agentic-flow/tree/main/agentic-flow/src/router) |
 | **QUIC Transport** | Ultra-low latency agent communication via Rust/WASM QUIC protocol | 50-70% faster than TCP, 0-RTT | [Docs](https://github.com/ruvnet/agentic-flow/tree/main/crates/agentic-flow-quic) |
 
-**CLI Usage**:
-- **AgentDB**: Full CLI with 17 commands (`npx agentdb <command>`)
-- **Multi-Model Router**: Via `--optimize` flag
-- **Agent Booster**: Automatic on code edits
-- **ReasoningBank**: API only
-- **QUIC Transport**: API only
-
-**Programmatic**: All components importable: `agentic-flow/agentdb`, `agentic-flow/router`, `agentic-flow/reasoningbank`, `agentic-flow/agent-booster`, `agentic-flow/transport/quic`
+**CLI Usage**: Multi-Model Router via `--optimize`, Agent Booster (automatic), ReasoningBank (API only), QUIC Transport (API only)
+**Programmatic**: All components importable: `agentic-flow/router`, `agentic-flow/reasoningbank`, `agentic-flow/agent-booster`, `agentic-flow/transport/quic`
 
 **Get Started:**
 ```bash
-# CLI: AgentDB memory operations
-npx agentdb reflexion store "session-1" "implement_auth" 0.95 true "Success!"
-npx agentdb skill search "authentication" 10
-npx agentdb causal query "" "code_quality" 0.8
-npx agentdb learner run
-
 # CLI: Auto-optimization (Agent Booster runs automatically on code edits)
 npx agentic-flow --agent coder --task "Build a REST API" --optimize
 
 # Programmatic: Import any component
-import { ReflexionMemory, SkillLibrary, CausalMemoryGraph } from 'agentic-flow/agentdb';
 import { ModelRouter } from 'agentic-flow/router';
 import * as reasoningbank from 'agentic-flow/reasoningbank';
 import { AgentBooster } from 'agentic-flow/agent-booster';
@@ -262,10 +248,18 @@ npx agentic-flow mcp status  # Check server status
 npx agentic-flow --list              # List all 79 agents
 npx agentic-flow agent info coder    # Get agent details
 npx agentic-flow agent create        # Create custom agent
+
+# Skills management (NEW in v1.5.11+)
+npx agentic-flow skills list         # List all installed skills
+npx agentic-flow skills init         # Initialize skills directories
+npx agentic-flow skills create       # Create agentic-flow example skills
+npx agentic-flow skills init-builder # Install skill-builder framework
+npx agentic-flow skills help         # Show skills help
 ```
 
 **Built-in MCP Tools** (7): agent execution, list agents, create agent, agent info, conflicts check, model optimizer, list all agents
 **External MCP Servers**: claude-flow (101 tools), flow-nexus (96 tools), agentic-payments (10 tools)
+**Skills System**: Create reusable AI workflows with YAML frontmatter and progressive disclosure
 
 ---
 
@@ -381,6 +375,76 @@ for (const file of files) {
 
 ---
 
+## 🎓 Skills System (Claude Code Integration)
+
+**NEW in v1.5.11+**: Create reusable AI workflows with proper YAML frontmatter, progressive disclosure, and Claude Code integration.
+
+### What Are Skills?
+
+Skills are structured AI workflows that Claude can autonomously discover and execute across all surfaces (Claude.ai, Claude Code, SDK, API). Each skill is a self-contained SKILL.md file with YAML frontmatter defining its capabilities.
+
+### Quick Start
+
+```bash
+# Initialize skills directories
+npx agentic-flow skills init         # Creates ~/.claude/skills and .claude/skills
+
+# Install skill-builder framework (for creating custom skills)
+npx agentic-flow skills init-builder # Adds skill-builder to your project
+
+# Create example agentic-flow skills
+npx agentic-flow skills create       # Creates AgentDB, swarm, reasoningbank skills
+
+# List all installed skills
+npx agentic-flow skills list         # Shows personal and project skills
+
+# Get help
+npx agentic-flow skills help         # Show all commands and usage
+```
+
+### Skill Structure
+
+Every skill requires a SKILL.md file with YAML frontmatter:
+
+```yaml
+---
+name: "Skill Name"                    # REQUIRED: Max 64 chars
+description: "What this skill does    # REQUIRED: Max 1024 chars
+and when Claude should use it."       # Include BOTH what & when
+---
+
+# Skill Name
+
+## What This Skill Does
+[Instructions for Claude]
+
+## Quick Start
+[Basic usage examples]
+
+## Prerequisites
+[Requirements]
+```
+
+### Built-in Skills
+
+- **skill-builder** - Create new Claude Code Skills with proper structure
+- **agentdb-memory-patterns** - Persistent memory patterns for AI agents
+- **agentdb-vector-search** - Semantic search with AgentDB
+- **reasoningbank-intelligence** - Adaptive learning and pattern recognition
+- **swarm-orchestration** - Multi-agent coordination workflows
+
+### Benefits
+
+- ✅ **Reusable**: Share skills across projects and teams
+- ✅ **Discoverable**: Claude finds and uses skills automatically
+- ✅ **Structured**: YAML frontmatter ensures consistency
+- ✅ **Progressive**: Expand details only when needed
+- ✅ **Validated**: Built-in validation catches errors early
+
+**Learn More:** [Skills Documentation](https://github.com/ruvnet/agentic-flow/tree/main/agentic-flow/.claude/skills)
+
+---
+
 ## 🎛️ Programmatic API
 
 ### Multi-Model Router

package/dist/cli-proxy.js

@@ -31,8 +31,10 @@ import { parseArgs } from "./utils/cli.js";
 import { getAgent, listAgents } from "./utils/agentLoader.js";
 import { claudeAgent } from "./agents/claudeAgent.js";
 import { handleReasoningBankCommand } from "./utils/reasoningbankCommands.js";
+import { handleAgentDBCommand } from "./utils/agentdbCommands.js";
 import { handleConfigCommand } from "./cli/config-wizard.js";
 import { handleAgentCommand } from "./cli/agent-manager.js";
+import { handleSkillsCommand } from "./cli/skills-manager.js";
 import { ModelOptimizer } from "./utils/modelOptimizer.js";
 import { detectModelCapabilities } from "./utils/modelCapabilities.js";
 import { AgentBoosterPreprocessor } from "./utils/agentBoosterPreprocessor.js";
@@ -54,7 +56,7 @@ class AgenticFlowCLI {
             process.exit(0);
         }
         // If no mode and no agent specified, show help
-        if (!options.agent && options.mode !== 'list' && !['config', 'agent-manager', 'mcp-manager', 'proxy', 'quic', 'claude-code', 'mcp', 'reasoningbank'].includes(options.mode)) {
+        if (!options.agent && options.mode !== 'list' && !['config', 'agent-manager', 'mcp-manager', 'agentdb', 'skills', 'proxy', 'quic', 'claude-code', 'mcp', 'reasoningbank'].includes(options.mode)) {
             this.printHelp();
             process.exit(0);
         }
@@ -74,6 +76,18 @@ class AgenticFlowCLI {
             await handleAgentCommand(agentArgs);
             process.exit(0);
         }
+        if (options.mode === 'agentdb') {
+            // Handle AgentDB commands
+            const agentdbArgs = process.argv.slice(3); // Skip 'node', 'cli-proxy.js', 'agentdb'
+            await handleAgentDBCommand(agentdbArgs);
+            process.exit(0);
+        }
+        if (options.mode === 'skills') {
+            // Handle Skills commands
+            const skillsArgs = process.argv.slice(3); // Skip 'node', 'cli-proxy.js', 'skills'
+            await handleSkillsCommand(skillsArgs);
+            process.exit(0);
+        }
         if (options.mode === 'mcp-manager') {
             // Handle MCP manager commands (add, list, remove, etc.)
             const { spawn } = await import('child_process');
@@ -892,6 +906,7 @@ COMMANDS:
   config [subcommand]     Manage environment configuration (interactive wizard)
   mcp <command> [server]  Manage MCP servers (start, stop, status, list)
   agent <command>         Agent management (list, create, info, conflicts)
+  agentdb <command>       AgentDB vector database management (init, search, migrate, etc.)
   proxy [options]         Run standalone proxy server for Claude Code/Cursor
   quic [options]          Run QUIC transport proxy for ultra-low latency (50-70% faster)
   claude-code [options]   Spawn Claude Code with auto-configured proxy
@@ -920,6 +935,21 @@ AGENT COMMANDS:
   npx agentic-flow agent info <name>     Show detailed agent information
   npx agentic-flow agent conflicts       Check for package/local conflicts
 
+AGENTDB COMMANDS (Vector Database for ReasoningBank):
+  npx agentic-flow agentdb init          Initialize AgentDB database
+  npx agentic-flow agentdb search        Search similar patterns (vector similarity)
+  npx agentic-flow agentdb insert        Insert pattern with embedding
+  npx agentic-flow agentdb train         Train learning model on experiences
+  npx agentic-flow agentdb stats         Display database statistics
+  npx agentic-flow agentdb optimize      Optimize database (consolidation, pruning)
+  npx agentic-flow agentdb migrate       Migrate from legacy ReasoningBank
+  npx agentic-flow agentdb export        Export patterns to JSON
+  npx agentic-flow agentdb import        Import patterns from JSON
+  npx agentic-flow agentdb help          Show detailed AgentDB help
+
+  Performance: 150x-12,500x faster than legacy ReasoningBank
+  Features: HNSW indexing, learning plugins, reasoning agents, QUIC sync
+
 OPTIONS:
   --task, -t <task>           Task description for agent mode
   --model, -m <model>         Model to use (triggers OpenRouter if contains "/")
@@ -1018,11 +1048,12 @@ OPENROUTER MODELS (Best Free Tested):
   All models above support OpenRouter leaderboard tracking via HTTP-Referer headers.
   See https://openrouter.ai/models for full model catalog.
 
-MCP TOOLS (213+ available):
+MCP TOOLS (223+ available):
   • agentic-flow: 7 tools (agent execution, creation, management, model optimization)
   • claude-flow: 101 tools (neural networks, GitHub, workflows, DAA)
   • flow-nexus: 96 cloud tools (sandboxes, distributed swarms, templates)
   • agentic-payments: 6 tools (payment authorization, multi-agent consensus)
+  • agentdb: 10 tools (vector search, learning, reasoning, optimization)
 
 OPTIMIZATION BENEFITS:
   💰 Cost Savings: 85-98% cheaper models for same quality tasks

package/dist/mcp/standalone-stdio.js

@@ -533,198 +533,7 @@ server.addTool({
         }
     }
 });
-// ========================================
-// AgentDB Vector Database Tools (Core 5)
-// ========================================
-// Tool: Get database statistics
-server.addTool({
-    name: 'agentdb_stats',
-    description: 'Enhanced database statistics including table counts, memory usage, and performance metrics for all AgentDB tables.',
-    parameters: z.object({}),
-    execute: async () => {
-        try {
-            const cmd = `npx agentdb db stats`;
-            const result = execSync(cmd, {
-                encoding: 'utf-8',
-                maxBuffer: 5 * 1024 * 1024,
-                timeout: 10000
-            });
-            return JSON.stringify({
-                success: true,
-                stats: result.trim(),
-                timestamp: new Date().toISOString()
-            }, null, 2);
-        }
-        catch (error) {
-            throw new Error(`Failed to get database stats: ${error.message}`);
-        }
-    }
-});
-// Tool: Store reasoning pattern
-server.addTool({
-    name: 'agentdb_pattern_store',
-    description: 'Store a reasoning pattern in ReasoningBank for future retrieval and learning. Patterns capture successful problem-solving approaches.',
-    parameters: z.object({
-        sessionId: z.string().describe('Session identifier for the pattern'),
-        task: z.string().describe('Task description that was solved'),
-        reward: z.number().min(0).max(1).describe('Success metric (0-1, where 1 is perfect success)'),
-        success: z.boolean().describe('Whether the task was completed successfully'),
-        critique: z.string().optional().describe('Self-reflection or critique of the approach'),
-        input: z.string().optional().describe('Input or context for the task'),
-        output: z.string().optional().describe('Output or solution generated'),
-        latencyMs: z.number().optional().describe('Time taken in milliseconds'),
-        tokensUsed: z.number().optional().describe('Number of tokens consumed')
-    }),
-    execute: async ({ sessionId, task, reward, success, critique, input, output, latencyMs, tokensUsed }) => {
-        try {
-            const args = [
-                sessionId,
-                `"${task}"`,
-                reward.toString(),
-                success.toString()
-            ];
-            if (critique)
-                args.push(`"${critique}"`);
-            if (input)
-                args.push(`"${input}"`);
-            if (output)
-                args.push(`"${output}"`);
-            if (latencyMs !== undefined)
-                args.push(latencyMs.toString());
-            if (tokensUsed !== undefined)
-                args.push(tokensUsed.toString());
-            const cmd = `npx agentdb reflexion store ${args.join(' ')}`;
-            const result = execSync(cmd, {
-                encoding: 'utf-8',
-                maxBuffer: 10 * 1024 * 1024,
-                timeout: 30000
-            });
-            return JSON.stringify({
-                success: true,
-                sessionId,
-                task: task.substring(0, 100) + (task.length > 100 ? '...' : ''),
-                reward,
-                stored: true,
-                message: 'Pattern stored successfully in ReasoningBank',
-                output: result.trim()
-            }, null, 2);
-        }
-        catch (error) {
-            throw new Error(`Failed to store pattern: ${error.message}`);
-        }
-    }
-});
-// Tool: Search reasoning patterns
-server.addTool({
-    name: 'agentdb_pattern_search',
-    description: 'Search for similar reasoning patterns in ReasoningBank by task description. Retrieves past successful approaches to guide current problem-solving.',
-    parameters: z.object({
-        task: z.string().describe('Task description to search for similar patterns'),
-        k: z.number().optional().default(5).describe('Number of results to retrieve (default: 5)'),
-        minReward: z.number().optional().describe('Minimum reward threshold (0-1) to filter results'),
-        onlySuccesses: z.boolean().optional().describe('Only retrieve successful episodes'),
-        onlyFailures: z.boolean().optional().describe('Only retrieve failed episodes (for learning)')
-    }),
-    execute: async ({ task, k, minReward, onlySuccesses, onlyFailures }) => {
-        try {
-            const args = [`"${task}"`, (k || 5).toString()];
-            if (minReward !== undefined)
-                args.push(minReward.toString());
-            if (onlyFailures)
-                args.push('true', 'false');
-            else if (onlySuccesses)
-                args.push('false', 'true');
-            const cmd = `npx agentdb reflexion retrieve ${args.join(' ')}`;
-            const result = execSync(cmd, {
-                encoding: 'utf-8',
-                maxBuffer: 10 * 1024 * 1024,
-                timeout: 30000
-            });
-            return JSON.stringify({
-                success: true,
-                query: task,
-                k: k || 5,
-                filters: {
-                    minReward,
-                    onlySuccesses: onlySuccesses || false,
-                    onlyFailures: onlyFailures || false
-                },
-                results: result.trim(),
-                message: `Retrieved ${k || 5} similar patterns from ReasoningBank`
-            }, null, 2);
-        }
-        catch (error) {
-            throw new Error(`Failed to search patterns: ${error.message}`);
-        }
-    }
-});
-// Tool: Get pattern statistics
-server.addTool({
-    name: 'agentdb_pattern_stats',
-    description: 'Get aggregated statistics and critique summary for patterns related to a specific task. Provides insights from past attempts.',
-    parameters: z.object({
-        task: z.string().describe('Task description to analyze'),
-        k: z.number().optional().default(5).describe('Number of recent patterns to analyze (default: 5)')
-    }),
-    execute: async ({ task, k }) => {
-        try {
-            const cmd = `npx agentdb reflexion critique-summary "${task}" ${k || 5}`;
-            const result = execSync(cmd, {
-                encoding: 'utf-8',
-                maxBuffer: 10 * 1024 * 1024,
-                timeout: 30000
-            });
-            return JSON.stringify({
-                success: true,
-                task,
-                analyzedPatterns: k || 5,
-                summary: result.trim(),
-                message: 'Pattern statistics and critique summary generated'
-            }, null, 2);
-        }
-        catch (error) {
-            throw new Error(`Failed to get pattern stats: ${error.message}`);
-        }
-    }
-});
-// Tool: Clear query cache
-server.addTool({
-    name: 'agentdb_clear_cache',
-    description: 'Clear the AgentDB query cache to free memory or force fresh queries. Useful after bulk updates or when debugging.',
-    parameters: z.object({
-        confirm: z.boolean().optional().default(false).describe('Confirmation to clear cache (set to true to proceed)')
-    }),
-    execute: async ({ confirm }) => {
-        try {
-            if (!confirm) {
-                return JSON.stringify({
-                    success: false,
-                    message: 'Cache clear operation requires confirmation. Set confirm=true to proceed.',
-                    warning: 'Clearing cache will remove all cached query results and may temporarily impact performance.'
-                }, null, 2);
-            }
-            // Since there's no direct CLI command for cache clearing, we'll provide information
-            // In a real implementation, this would call a cache management function
-            const info = {
-                success: true,
-                operation: 'cache_clear',
-                message: 'Query cache cleared successfully',
-                note: 'AgentDB uses in-memory cache for query optimization. Cache will rebuild automatically on next queries.',
-                timestamp: new Date().toISOString(),
-                nextSteps: [
-                    'First queries after cache clear may be slower',
-                    'Cache will warm up with frequently accessed patterns',
-                    'Consider running agentdb_stats to verify memory reduction'
-                ]
-            };
-            return JSON.stringify(info, null, 2);
-        }
-        catch (error) {
-            throw new Error(`Failed to clear cache: ${error.message}`);
-        }
-    }
-});
-console.error('✅ Registered 15 tools (7 agentic-flow + 3 agent-booster + 5 agentdb):');
+console.error('✅ Registered 10 tools (7 agentic-flow + 3 agent-booster):');
 console.error('   • agentic_flow_agent (execute agent with 13 parameters)');
 console.error('   • agentic_flow_list_agents (list 66+ agents)');
 console.error('   • agentic_flow_create_agent (create custom agent)');
@@ -732,14 +541,9 @@ console.error('   • agentic_flow_list_all_agents (list with sources)');
 console.error('   • agentic_flow_agent_info (get agent details)');
 console.error('   • agentic_flow_check_conflicts (conflict detection)');
 console.error('   • agentic_flow_optimize_model (auto-select best model)');
-console.error('   • agent_booster_edit_file (352x faster code editing) ⚡');
-console.error('   • agent_booster_batch_edit (multi-file refactoring) ⚡');
-console.error('   • agent_booster_parse_markdown (LLM output parsing) ⚡');
-console.error('   • agentdb_stats (database statistics) 🧠 NEW');
-console.error('   • agentdb_pattern_store (store reasoning patterns) 🧠 NEW');
-console.error('   • agentdb_pattern_search (search similar patterns) 🧠 NEW');
-console.error('   • agentdb_pattern_stats (pattern analytics) 🧠 NEW');
-console.error('   • agentdb_clear_cache (clear query cache) 🧠 NEW');
+console.error('   • agent_booster_edit_file (352x faster code editing) ⚡ NEW');
+console.error('   • agent_booster_batch_edit (multi-file refactoring) ⚡ NEW');
+console.error('   • agent_booster_parse_markdown (LLM output parsing) ⚡ NEW');
 console.error('🔌 Starting stdio transport...');
 server.start({ transportType: 'stdio' }).then(() => {
     console.error('✅ Agentic-Flow MCP server running on stdio');

package/dist/reasoningbank/index.js

@@ -99,6 +99,10 @@ export async function runTask(options) {
         consolidated
     };
 }
+// AgentDB Integration
+// Ultra-fast vector database drop-in replacement
+// 150x-12,500x faster than legacy implementation
+export { createAgentDBAdapter, createDefaultAgentDBAdapter, migrateToAgentDB, validateMigration } from './agentdb-adapter.js';
 // Version info
 export const VERSION = '1.0.0';
 export const PAPER_URL = 'https://arxiv.org/html/2509.25140v1';

package/dist/utils/cli.js

@@ -49,6 +49,16 @@ export function parseArgs() {
         options.mode = 'reasoningbank';
         return options;
     }
+    // Check for agentdb command
+    if (args[0] === 'agentdb') {
+        options.mode = 'agentdb';
+        return options;
+    }
+    // Check for skills command
+    if (args[0] === 'skills') {
+        options.mode = 'skills';
+        return options;
+    }
     for (let i = 0; i < args.length; i++) {
         const arg = args[i];
         switch (arg) {
@@ -155,6 +165,7 @@ USAGE:
   npx agentic-flow [COMMAND] [OPTIONS]
 
 COMMANDS:
+  skills <command>        Claude Code Skills management (init, create, list)
   reasoningbank <cmd>     Memory system that learns from experience (demo, test, init)
   claude-code [options]   Spawn Claude Code with proxy + Agent Booster (57x faster edits)
   mcp <command> [server]  Manage MCP servers (start, stop, status, list)
@@ -171,6 +182,12 @@ REASONINGBANK COMMANDS:
   npx agentic-flow reasoningbank benchmark    Run performance benchmarks
   npx agentic-flow reasoningbank status       Show memory statistics
 
+SKILLS COMMANDS:
+  npx agentic-flow skills init [location]   Initialize skills directories (personal/project/both)
+  npx agentic-flow skills create            Create example agentic-flow skills
+  npx agentic-flow skills list              List all installed skills
+  npx agentic-flow skills help              Show skills help
+
 MCP COMMANDS:
   npx agentic-flow mcp start [server]    Start MCP server(s)
   npx agentic-flow mcp stop [server]     Stop MCP server(s)
@@ -216,6 +233,11 @@ OPTIONS:
   --help, -h                  Show this help message
 
 EXAMPLES:
+  # Skills (Claude Code integration!)
+  npx agentic-flow skills init              # Initialize skills directories
+  npx agentic-flow skills create            # Create agentdb-quickstart skill
+  npx agentic-flow skills list              # List all installed skills
+
   # ReasoningBank (Learn from agent experience!)
   npx agentic-flow reasoningbank demo       # See 0% → 100% success transformation
   npx agentic-flow reasoningbank test       # Run 27 validation tests