[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7BP3BFQGCYCHR2REBNA561G":3,"skill-same:01M7BP3BFQGCYCHR2REBNA561G":47},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":24,"commitSha":25,"skillDirectory":26,"forksCount":27,"repositoryHtmlUrl":28,"repositoryHomepageUrl":29,"ownerHtmlUrl":30,"ownerBlogUrl":6,"ownerDescription":31,"updatedAt":32,"install":33,"resources":46},"01M7BP3BFQGCYCHR2REBNA561G","AgentDB Learning Plugins",null,"Create and train AI learning plugins with AgentDB's 9 reinforcement learning algorithms. Includes Decision Transformer, Q-Learning, SARSA, Actor-Critic, and more. Use when building self-learning agents, implementing RL, or optimizing agent behavior through experience.","YYC-Cube\u002FYYC3-Knowledge-Base","YYC-Cube","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F235006543?v=4","01M79J2PH4SWD2X0825X6PN748","yyc3","01M79J2PHVK3F6ZBYT8K2SCZCW",1,"MIT","https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-Agent-智能体\u002FYYC3-Agent-智能体-ruflo\u002Fv3\u002F@claude-flow\u002Fcli\u002F.claude\u002Fskills\u002Fagentdb-learning\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: \"AgentDB Learning Plugins\"\ndescription: \"Create and train AI learning plugins with AgentDB's 9 reinforcement learning algorithms. Includes Decision Transformer, Q-Learning, SARSA, Actor-Critic, and more. Use when building self-learning agents, implementing RL, or optimizing agent behavior through experience.\"\n---\n\n# AgentDB Learning Plugins\n\n## What This Skill Does\n\nProvides access to 9 reinforcement learning algorithms via AgentDB's plugin system. Create, train, and deploy learning plugins for autonomous agents that improve through experience. Includes offline RL (Decision Transformer), value-based learning (Q-Learning), policy gradients (Actor-Critic), and advanced techniques.\n\n**Performance**: Train models 10-100x faster with WASM-accelerated neural inference.\n\n## Prerequisites\n\n- Node.js 18+\n- AgentDB v1.0.7+ (via agentic-flow)\n- Basic understanding of reinforcement learning (recommended)\n\n---\n\n## Quick Start with CLI\n\n### Create Learning Plugin\n\n```bash\n# Interactive wizard\nnpx agentdb@latest create-plugin\n\n# Use specific template\nnpx agentdb@latest create-plugin -t decision-transformer -n my-agent\n\n# Preview without creating\nnpx agentdb@latest create-plugin -t q-learning --dry-run\n\n# Custom output directory\nnpx agentdb@latest create-plugin -t actor-critic -o .\u002Fplugins\n```\n\n### List Available Templates\n\n```bash\n# Show all plugin templates\nnpx agentdb@latest list-templates\n\n# Available templates:\n# - decision-transformer (sequence modeling RL - recommended)\n# - q-learning (value-based learning)\n# - sarsa (on-policy TD learning)\n# - actor-critic (policy gradient with baseline)\n# - curiosity-driven (exploration-based)\n```\n\n### Manage Plugins\n\n```bash\n# List installed plugins\nnpx agentdb@latest list-plugins\n\n# Get plugin information\nnpx agentdb@latest plugin-info my-agent\n\n# Shows: algorithm, configuration, training status\n```\n\n---\n\n## Quick Start with API\n\n```typescript\nimport { createAgentDBAdapter } from 'agentic-flow\u002Freasoningbank';\n\n\u002F\u002F Initialize with learning enabled\nconst adapter = await createAgentDBAdapter({\n  dbPath: '.agentdb\u002Flearning.db',\n  enableLearning: true,       \u002F\u002F Enable learning plugins\n  enableReasoning: true,\n  cacheSize: 1000,\n});\n\n\u002F\u002F Store training experience\nawait adapter.insertPattern({\n  id: '',\n  type: 'experience',\n  domain: 'game-playing',\n  pattern_data: JSON.stringify({\n    embedding: await computeEmbedding('state-action-reward'),\n    pattern: {\n      state: [0.1, 0.2, 0.3],\n      action: 2,\n      reward: 1.0,\n      next_state: [0.15, 0.25, 0.35],\n      done: false\n    }\n  }),\n  confidence: 0.9,\n  usage_count: 1,\n  success_count: 1,\n  created_at: Date.now(),\n  last_used: Date.now(),\n});\n\n\u002F\u002F Train learning model\nconst metrics = await adapter.train({\n  epochs: 50,\n  batchSize: 32,\n});\n\nconsole.log('Training Loss:', metrics.loss);\nconsole.log('Duration:', metrics.duration, 'ms');\n```\n\n---\n\n## Available Learning Algorithms (9 Total)\n\n### 1. Decision Transformer (Recommended)\n\n**Type**: Offline Reinforcement Learning\n**Best For**: Learning from logged experiences, imitation learning\n**Strengths**: No online interaction needed, stable training\n\n```bash\nnpx agentdb@latest create-plugin -t decision-transformer -n dt-agent\n```\n\n**Use Cases**:\n- Learn from historical data\n- Imitation learning from expert demonstrations\n- Safe learning without environment interaction\n- Sequence modeling tasks\n\n**Configuration**:\n```json\n{\n  \"algorithm\": \"decision-transformer\",\n  \"model_size\": \"base\",\n  \"context_length\": 20,\n  \"embed_dim\": 128,\n  \"n_heads\": 8,\n  \"n_layers\": 6\n}\n```\n\n### 2. Q-Learning\n\n**Type**: Value-Based RL (Off-Policy)\n**Best For**: Discrete action spaces, sample efficiency\n**Strengths**: Proven, simple, works well for small\u002Fmedium problems\n\n```bash\nnpx agentdb@latest create-plugin -t q-learning -n q-agent\n```\n\n**Use Cases**:\n- Grid worlds, board games\n- Navigation tasks\n- Resource allocation\n- Discrete decision-making\n\n**Configuration**:\n```json\n{\n  \"algorithm\": \"q-learning\",\n  \"learning_rate\": 0.001,\n  \"gamma\": 0.99,\n  \"epsilon\": 0.1,\n  \"epsilon_decay\": 0.995\n}\n```\n\n### 3. SARSA\n\n**Type**: Value-Based RL (On-Policy)\n**Best For**: Safe exploration, risk-sensitive tasks\n**Strengths**: More conservative than Q-Learning, better for safety\n\n```bash\nnpx agentdb@latest create-plugin -t sarsa -n sarsa-agent\n```\n\n**Use Cases**:\n- Safety-critical applications\n- Risk-sensitive decision-making\n- Online learning with exploration\n\n**Configuration**:\n```json\n{\n  \"algorithm\": \"sarsa\",\n  \"learning_rate\": 0.001,\n  \"gamma\": 0.99,\n  \"epsilon\": 0.1\n}\n```\n\n### 4. Actor-Critic\n\n**Type**: Policy Gradient with Value Baseline\n**Best For**: Continuous actions, variance reduction\n**Strengths**: Stable, works for continuous\u002Fdiscrete actions\n\n```bash\nnpx agentdb@latest create-plugin -t actor-critic -n ac-agent\n```\n\n**Use Cases**:\n- Continuous control (robotics, simulations)\n- Complex action spaces\n- Multi-agent coordination\n\n**Configuration**:\n```json\n{\n  \"algorithm\": \"actor-critic\",\n  \"actor_lr\": 0.001,\n  \"critic_lr\": 0.002,\n  \"gamma\": 0.99,\n  \"entropy_coef\": 0.01\n}\n```\n\n### 5. Active Learning\n\n**Type**: Query-Based Learning\n**Best For**: Label-efficient learning, human-in-the-loop\n**Strengths**: Minimizes labeling cost, focuses on uncertain samples\n\n**Use Cases**:\n- Human feedback incorporation\n- Label-efficient training\n- Uncertainty sampling\n- Annotation cost reduction\n\n### 6. Adversarial Training\n\n**Type**: Robustness Enhancement\n**Best For**: Safety, robustness to perturbations\n**Strengths**: Improves model robustness, adversarial defense\n\n**Use Cases**:\n- Security applications\n- Robust decision-making\n- Adversarial defense\n- Safety testing\n\n### 7. Curriculum Learning\n\n**Type**: Progressive Difficulty Training\n**Best For**: Complex tasks, faster convergence\n**Strengths**: Stable learning, faster convergence on hard tasks\n\n**Use Cases**:\n- Complex multi-stage tasks\n- Hard exploration problems\n- Skill composition\n- Transfer learning\n\n### 8. Federated Learning\n\n**Type**: Distributed Learning\n**Best For**: Privacy, distributed data\n**Strengths**: Privacy-preserving, scalable\n\n**Use Cases**:\n- Multi-agent systems\n- Privacy-sensitive data\n- Distributed training\n- Collaborative learning\n\n### 9. Multi-Task Learning\n\n**Type**: Transfer Learning\n**Best For**: Related tasks, knowledge sharing\n**Strengths**: Faster learning on new tasks, better generalization\n\n**Use Cases**:\n- Task families\n- Transfer learning\n- Domain adaptation\n- Meta-learning\n\n---\n\n## Training Workflow\n\n### 1. Collect Experiences\n\n```typescript\n\u002F\u002F Store experiences during agent execution\nfor (let i = 0; i \u003C numEpisodes; i++) {\n  const episode = runEpisode();\n\n  for (const step of episode.steps) {\n    await adapter.insertPattern({\n      id: '',\n      type: 'experience',\n      domain: 'task-domain',\n      pattern_data: JSON.stringify({\n        embedding: await computeEmbedding(JSON.stringify(step)),\n        pattern: {\n          state: step.state,\n          action: step.action,\n          reward: step.reward,\n          next_state: step.next_state,\n          done: step.done\n        }\n      }),\n      confidence: step.reward > 0 ? 0.9 : 0.5,\n      usage_count: 1,\n      success_count: step.reward > 0 ? 1 : 0,\n      created_at: Date.now(),\n      last_used: Date.now(),\n    });\n  }\n}\n```\n\n### 2. Train Model\n\n```typescript\n\u002F\u002F Train on collected experiences\nconst trainingMetrics = await adapter.train({\n  epochs: 100,\n  batchSize: 64,\n  learningRate: 0.001,\n  validationSplit: 0.2,\n});\n\nconsole.log('Training Metrics:', trainingMetrics);\n\u002F\u002F {\n\u002F\u002F   loss: 0.023,\n\u002F\u002F   valLoss: 0.028,\n\u002F\u002F   duration: 1523,\n\u002F\u002F   epochs: 100\n\u002F\u002F }\n```\n\n### 3. Evaluate Performance\n\n```typescript\n\u002F\u002F Retrieve similar successful experiences\nconst testQuery = await computeEmbedding(JSON.stringify(testState));\nconst result = await adapter.retrieveWithReasoning(testQuery, {\n  domain: 'task-domain',\n  k: 10,\n  synthesizeContext: true,\n});\n\n\u002F\u002F Evaluate action quality\nconst suggestedAction = result.memories[0].pattern.action;\nconst confidence = result.memories[0].similarity;\n\nconsole.log('Suggested Action:', suggestedAction);\nconsole.log('Confidence:', confidence);\n```\n\n---\n\n## Advanced Training Techniques\n\n### Experience Replay\n\n```typescript\n\u002F\u002F Store experiences in buffer\nconst replayBuffer = [];\n\n\u002F\u002F Sample random batch for training\nconst batch = sampleRandomBatch(replayBuffer, batchSize: 32);\n\n\u002F\u002F Train on batch\nawait adapter.train({\n  data: batch,\n  epochs: 1,\n  batchSize: 32,\n});\n```\n\n### Prioritized Experience Replay\n\n```typescript\n\u002F\u002F Store experiences with priority (TD error)\nawait adapter.insertPattern({\n  \u002F\u002F ... standard fields\n  confidence: tdError,  \u002F\u002F Use TD error as confidence\u002Fpriority\n  \u002F\u002F ...\n});\n\n\u002F\u002F Retrieve high-priority experiences\nconst highPriority = await adapter.retrieveWithReasoning(queryEmbedding, {\n  domain: 'task-domain',\n  k: 32,\n  minConfidence: 0.7,  \u002F\u002F Only high TD-error experiences\n});\n```\n\n### Multi-Agent Training\n\n```typescript\n\u002F\u002F Collect experiences from multiple agents\nfor (const agent of agents) {\n  const experience = await agent.step();\n\n  await adapter.insertPattern({\n    \u002F\u002F ... store experience with agent ID\n    domain: `multi-agent\u002F${agent.id}`,\n  });\n}\n\n\u002F\u002F Train shared model\nawait adapter.train({\n  epochs: 50,\n  batchSize: 64,\n});\n```\n\n---\n\n## Performance Optimization\n\n### Batch Training\n\n```typescript\n\u002F\u002F Collect batch of experiences\nconst experiences = collectBatch(size: 1000);\n\n\u002F\u002F Batch insert (500x faster)\nfor (const exp of experiences) {\n  await adapter.insertPattern({ \u002F* ... *\u002F });\n}\n\n\u002F\u002F Train on batch\nawait adapter.train({\n  epochs: 10,\n  batchSize: 128,  \u002F\u002F Larger batch for efficiency\n});\n```\n\n### Incremental Learning\n\n```typescript\n\u002F\u002F Train incrementally as new data arrives\nsetInterval(async () => {\n  const newExperiences = getNewExperiences();\n\n  if (newExperiences.length > 100) {\n    await adapter.train({\n      epochs: 5,\n      batchSize: 32,\n    });\n  }\n}, 60000);  \u002F\u002F Every minute\n```\n\n---\n\n## Integration with Reasoning Agents\n\nCombine learning with reasoning for better performance:\n\n```typescript\n\u002F\u002F Train learning model\nawait adapter.train({ epochs: 50, batchSize: 32 });\n\n\u002F\u002F Use reasoning agents for inference\nconst result = await adapter.retrieveWithReasoning(queryEmbedding, {\n  domain: 'decision-making',\n  k: 10,\n  useMMR: true,              \u002F\u002F Diverse experiences\n  synthesizeContext: true,    \u002F\u002F Rich context\n  optimizeMemory: true,       \u002F\u002F Consolidate patterns\n});\n\n\u002F\u002F Make decision based on learned experiences + reasoning\nconst decision = result.context.suggestedAction;\nconst confidence = result.memories[0].similarity;\n```\n\n---\n\n## CLI Operations\n\n```bash\n# Create plugin\nnpx agentdb@latest create-plugin -t decision-transformer -n my-plugin\n\n# List plugins\nnpx agentdb@latest list-plugins\n\n# Get plugin info\nnpx agentdb@latest plugin-info my-plugin\n\n# List templates\nnpx agentdb@latest list-templates\n```\n\n---\n\n## Troubleshooting\n\n### Issue: Training not converging\n```typescript\n\u002F\u002F Reduce learning rate\nawait adapter.train({\n  epochs: 100,\n  batchSize: 32,\n  learningRate: 0.0001,  \u002F\u002F Lower learning rate\n});\n```\n\n### Issue: Overfitting\n```typescript\n\u002F\u002F Use validation split\nawait adapter.train({\n  epochs: 50,\n  batchSize: 64,\n  validationSplit: 0.2,  \u002F\u002F 20% validation\n});\n\n\u002F\u002F Enable memory optimization\nawait adapter.retrieveWithReasoning(queryEmbedding, {\n  optimizeMemory: true,  \u002F\u002F Consolidate, reduce overfitting\n});\n```\n\n### Issue: Slow training\n```bash\n# Enable quantization for faster inference\n# Use binary quantization (32x faster)\n```\n\n---\n\n## Learn More\n\n- **Algorithm Papers**: See docs\u002Falgorithms\u002F for detailed papers\n- **GitHub**: https:\u002F\u002Fgithub.com\u002Fruvnet\u002Fagentic-flow\u002Ftree\u002Fmain\u002Fpackages\u002Fagentdb\n- **MCP Integration**: `npx agentdb@latest mcp`\n- **Website**: https:\u002F\u002Fagentdb.ruv.io\n\n---\n\n**Category**: Machine Learning \u002F Reinforcement Learning\n**Difficulty**: Intermediate to Advanced\n**Estimated Time**: 30-60 minutes\n","# AgentDB Learning Plugins\n\n## What This Skill Does\n\nProvides access to 9 reinforcement learning algorithms via AgentDB's plugin system. Create, train, and deploy learning plugins for autonomous agents that improve through experience. Includes offline RL (Decision Transformer), value-based learning (Q-Learning), policy gradients (Actor-Critic), and advanced techniques.\n\n**Performance**: Train models 10-100x faster with WASM-accelerated neural inference.\n\n## Prerequisites\n\n- Node.js 18+\n- AgentDB v1.0.7+ (via agentic-flow)\n- Basic understanding of reinforcement learning (recommended)\n\n---\n\n## Quick Start with CLI\n\n### Create Learning Plugin\n\n```bash\n# Interactive wizard\nnpx agentdb@latest create-plugin\n\n# Use specific template\nnpx agentdb@latest create-plugin -t decision-transformer -n my-agent\n\n# Preview without creating\nnpx agentdb@latest create-plugin -t q-learning --dry-run\n\n# Custom output directory\nnpx agentdb@latest create-plugin -t actor-critic -o .\u002Fplugins\n```\n\n### List Available Templates\n\n```bash\n# Show all plugin templates\nnpx agentdb@latest list-templates\n\n# Available templates:\n# - decision-transformer (sequence modeling RL - recommended)\n# - q-learning (value-based learning)\n# - sarsa (on-policy TD learning)\n# - actor-critic (policy gradient with baseline)\n# - curiosity-driven (exploration-based)\n```\n\n### Manage Plugins\n\n```bash\n# List installed plugins\nnpx agentdb@latest list-plugins\n\n# Get plugin information\nnpx agentdb@latest plugin-info my-agent\n\n# Shows: algorithm, configuration, training status\n```\n\n---\n\n## Quick Start with API\n\n```typescript\nimport { createAgentDBAdapter } from 'agentic-flow\u002Freasoningbank';\n\n\u002F\u002F Initialize with learning enabled\nconst adapter = await createAgentDBAdapter({\n  dbPath: '.agentdb\u002Flearning.db',\n  enableLearning: true,       \u002F\u002F Enable learning plugins\n  enableReasoning: true,\n  cacheSize: 1000,\n});\n\n\u002F\u002F Store training experience\nawait adapter.insertPattern({\n  id: '',\n  type: 'experience',\n  domain: 'game-playing',\n  pattern_data: JSON.stringify({\n    embedding: await computeEmbedding('state-action-reward'),\n    pattern: {\n      state: [0.1, 0.2, 0.3],\n      action: 2,\n      reward: 1.0,\n      next_state: [0.15, 0.25, 0.35],\n      done: false\n    }\n  }),\n  confidence: 0.9,\n  usage_count: 1,\n  success_count: 1,\n  created_at: Date.now(),\n  last_used: Date.now(),\n});\n\n\u002F\u002F Train learning model\nconst metrics = await adapter.train({\n  epochs: 50,\n  batchSize: 32,\n});\n\nconsole.log('Training Loss:', metrics.loss);\nconsole.log('Duration:', metrics.duration, 'ms');\n```\n\n---\n\n## Available Learning Algorithms (9 Total)\n\n### 1. Decision Transformer (Recommended)\n\n**Type**: Offline Reinforcement Learning\n**Best For**: Learning from logged experiences, imitation learning\n**Strengths**: No online interaction needed, stable training\n\n```bash\nnpx agentdb@latest create-plugin -t decision-transformer -n dt-agent\n```\n\n**Use Cases**:\n- Learn from historical data\n- Imitation learning from expert demonstrations\n- Safe learning without environment interaction\n- Sequence modeling tasks\n\n**Configuration**:\n```json\n{\n  \"algorithm\": \"decision-transformer\",\n  \"model_size\": \"base\",\n  \"context_length\": 20,\n  \"embed_dim\": 128,\n  \"n_heads\": 8,\n  \"n_layers\": 6\n}\n```\n\n### 2. Q-Learning\n\n**Type**: Value-Based RL (Off-Policy)\n**Best For**: Discrete action spaces, sample efficiency\n**Strengths**: Proven, simple, works well for small\u002Fmedium problems\n\n```bash\nnpx agentdb@latest create-plugin -t q-learning -n q-agent\n```\n\n**Use Cases**:\n- Grid worlds, board games\n- Navigation tasks\n- Resource allocation\n- Discrete decision-making\n\n**Configuration**:\n```json\n{\n  \"algorithm\": \"q-learning\",\n  \"learning_rate\": 0.001,\n  \"gamma\": 0.99,\n  \"epsilon\": 0.1,\n  \"epsilon_decay\": 0.995\n}\n```\n\n### 3. SARSA\n\n**Type**: Value-Based RL (On-Policy)\n**Best For**: Safe exploration, risk-sensitive tasks\n**Strengths**: More conservative than Q-Learning, better for safety\n\n```bash\nnpx agentdb@latest create-plugin -t sarsa -n sarsa-agent\n```\n\n**Use Cases**:\n- Safety-critical applications\n- Risk-sensitive decision-making\n- Online learning with exploration\n\n**Configuration**:\n```json\n{\n  \"algorithm\": \"sarsa\",\n  \"learning_rate\": 0.001,\n  \"gamma\": 0.99,\n  \"epsilon\": 0.1\n}\n```\n\n### 4. Actor-Critic\n\n**Type**: Policy Gradient with Value Baseline\n**Best For**: Continuous actions, variance reduction\n**Strengths**: Stable, works for continuous\u002Fdiscrete actions\n\n```bash\nnpx agentdb@latest create-plugin -t actor-critic -n ac-agent\n```\n\n**Use Cases**:\n- Continuous control (robotics, simulations)\n- Complex action spaces\n- Multi-agent coordination\n\n**Configuration**:\n```json\n{\n  \"algorithm\": \"actor-critic\",\n  \"actor_lr\": 0.001,\n  \"critic_lr\": 0.002,\n  \"gamma\": 0.99,\n  \"entropy_coef\": 0.01\n}\n```\n\n### 5. Active Learning\n\n**Type**: Query-Based Learning\n**Best For**: Label-efficient learning, human-in-the-loop\n**Strengths**: Minimizes labeling cost, focuses on uncertain samples\n\n**Use Cases**:\n- Human feedback incorporation\n- Label-efficient training\n- Uncertainty sampling\n- Annotation cost reduction\n\n### 6. Adversarial Training\n\n**Type**: Robustness Enhancement\n**Best For**: Safety, robustness to perturbations\n**Strengths**: Improves model robustness, adversarial defense\n\n**Use Cases**:\n- Security applications\n- Robust decision-making\n- Adversarial defense\n- Safety testing\n\n### 7. Curriculum Learning\n\n**Type**: Progressive Difficulty Training\n**Best For**: Complex tasks, faster convergence\n**Strengths**: Stable learning, faster convergence on hard tasks\n\n**Use Cases**:\n- Complex multi-stage tasks\n- Hard exploration problems\n- Skill composition\n- Transfer learning\n\n### 8. Federated Learning\n\n**Type**: Distributed Learning\n**Best For**: Privacy, distributed data\n**Strengths**: Privacy-preserving, scalable\n\n**Use Cases**:\n- Multi-agent systems\n- Privacy-sensitive data\n- Distributed training\n- Collaborative learning\n\n### 9. Multi-Task Learning\n\n**Type**: Transfer Learning\n**Best For**: Related tasks, knowledge sharing\n**Strengths**: Faster learning on new tasks, better generalization\n\n**Use Cases**:\n- Task families\n- Transfer learning\n- Domain adaptation\n- Meta-learning\n\n---\n\n## Training Workflow\n\n### 1. Collect Experiences\n\n```typescript\n\u002F\u002F Store experiences during agent execution\nfor (let i = 0; i \u003C numEpisodes; i++) {\n  const episode = runEpisode();\n\n  for (const step of episode.steps) {\n    await adapter.insertPattern({\n      id: '',\n      type: 'experience',\n      domain: 'task-domain',\n      pattern_data: JSON.stringify({\n        embedding: await computeEmbedding(JSON.stringify(step)),\n        pattern: {\n          state: step.state,\n          action: step.action,\n          reward: step.reward,\n          next_state: step.next_state,\n          done: step.done\n        }\n      }),\n      confidence: step.reward > 0 ? 0.9 : 0.5,\n      usage_count: 1,\n      success_count: step.reward > 0 ? 1 : 0,\n      created_at: Date.now(),\n      last_used: Date.now(),\n    });\n  }\n}\n```\n\n### 2. Train Model\n\n```typescript\n\u002F\u002F Train on collected experiences\nconst trainingMetrics = await adapter.train({\n  epochs: 100,\n  batchSize: 64,\n  learningRate: 0.001,\n  validationSplit: 0.2,\n});\n\nconsole.log('Training Metrics:', trainingMetrics);\n\u002F\u002F {\n\u002F\u002F   loss: 0.023,\n\u002F\u002F   valLoss: 0.028,\n\u002F\u002F   duration: 1523,\n\u002F\u002F   epochs: 100\n\u002F\u002F }\n```\n\n### 3. Evaluate Performance\n\n```typescript\n\u002F\u002F Retrieve similar successful experiences\nconst testQuery = await computeEmbedding(JSON.stringify(testState));\nconst result = await adapter.retrieveWithReasoning(testQuery, {\n  domain: 'task-domain',\n  k: 10,\n  synthesizeContext: true,\n});\n\n\u002F\u002F Evaluate action quality\nconst suggestedAction = result.memories[0].pattern.action;\nconst confidence = result.memories[0].similarity;\n\nconsole.log('Suggested Action:', suggestedAction);\nconsole.log('Confidence:', confidence);\n```\n\n---\n\n## Advanced Training Techniques\n\n### Experience Replay\n\n```typescript\n\u002F\u002F Store experiences in buffer\nconst replayBuffer = [];\n\n\u002F\u002F Sample random batch for training\nconst batch = sampleRandomBatch(replayBuffer, batchSize: 32);\n\n\u002F\u002F Train on batch\nawait adapter.train({\n  data: batch,\n  epochs: 1,\n  batchSize: 32,\n});\n```\n\n### Prioritized Experience Replay\n\n```typescript\n\u002F\u002F Store experiences with priority (TD error)\nawait adapter.insertPattern({\n  \u002F\u002F ... standard fields\n  confidence: tdError,  \u002F\u002F Use TD error as confidence\u002Fpriority\n  \u002F\u002F ...\n});\n\n\u002F\u002F Retrieve high-priority experiences\nconst highPriority = await adapter.retrieveWithReasoning(queryEmbedding, {\n  domain: 'task-domain',\n  k: 32,\n  minConfidence: 0.7,  \u002F\u002F Only high TD-error experiences\n});\n```\n\n### Multi-Agent Training\n\n```typescript\n\u002F\u002F Collect experiences from multiple agents\nfor (const agent of agents) {\n  const experience = await agent.step();\n\n  await adapter.insertPattern({\n    \u002F\u002F ... store experience with agent ID\n    domain: `multi-agent\u002F${agent.id}`,\n  });\n}\n\n\u002F\u002F Train shared model\nawait adapter.train({\n  epochs: 50,\n  batchSize: 64,\n});\n```\n\n---\n\n## Performance Optimization\n\n### Batch Training\n\n```typescript\n\u002F\u002F Collect batch of experiences\nconst experiences = collectBatch(size: 1000);\n\n\u002F\u002F Batch insert (500x faster)\nfor (const exp of experiences) {\n  await adapter.insertPattern({ \u002F* ... *\u002F });\n}\n\n\u002F\u002F Train on batch\nawait adapter.train({\n  epochs: 10,\n  batchSize: 128,  \u002F\u002F Larger batch for efficiency\n});\n```\n\n### Incremental Learning\n\n```typescript\n\u002F\u002F Train incrementally as new data arrives\nsetInterval(async () => {\n  const newExperiences = getNewExperiences();\n\n  if (newExperiences.length > 100) {\n    await adapter.train({\n      epochs: 5,\n      batchSize: 32,\n    });\n  }\n}, 60000);  \u002F\u002F Every minute\n```\n\n---\n\n## Integration with Reasoning Agents\n\nCombine learning with reasoning for better performance:\n\n```typescript\n\u002F\u002F Train learning model\nawait adapter.train({ epochs: 50, batchSize: 32 });\n\n\u002F\u002F Use reasoning agents for inference\nconst result = await adapter.retrieveWithReasoning(queryEmbedding, {\n  domain: 'decision-making',\n  k: 10,\n  useMMR: true,              \u002F\u002F Diverse experiences\n  synthesizeContext: true,    \u002F\u002F Rich context\n  optimizeMemory: true,       \u002F\u002F Consolidate patterns\n});\n\n\u002F\u002F Make decision based on learned experiences + reasoning\nconst decision = result.context.suggestedAction;\nconst confidence = result.memories[0].similarity;\n```\n\n---\n\n## CLI Operations\n\n```bash\n# Create plugin\nnpx agentdb@latest create-plugin -t decision-transformer -n my-plugin\n\n# List plugins\nnpx agentdb@latest list-plugins\n\n# Get plugin info\nnpx agentdb@latest plugin-info my-plugin\n\n# List templates\nnpx agentdb@latest list-templates\n```\n\n---\n\n## Troubleshooting\n\n### Issue: Training not converging\n```typescript\n\u002F\u002F Reduce learning rate\nawait adapter.train({\n  epochs: 100,\n  batchSize: 32,\n  learningRate: 0.0001,  \u002F\u002F Lower learning rate\n});\n```\n\n### Issue: Overfitting\n```typescript\n\u002F\u002F Use validation split\nawait adapter.train({\n  epochs: 50,\n  batchSize: 64,\n  validationSplit: 0.2,  \u002F\u002F 20% validation\n});\n\n\u002F\u002F Enable memory optimization\nawait adapter.retrieveWithReasoning(queryEmbedding, {\n  optimizeMemory: true,  \u002F\u002F Consolidate, reduce overfitting\n});\n```\n\n### Issue: Slow training\n```bash\n# Enable quantization for faster inference\n# Use binary quantization (32x faster)\n```\n\n---\n\n## Learn More\n\n- **Algorithm Papers**: See docs\u002Falgorithms\u002F for detailed papers\n- **GitHub**: https:\u002F\u002Fgithub.com\u002Fruvnet\u002Fagentic-flow\u002Ftree\u002Fmain\u002Fpackages\u002Fagentdb\n- **MCP Integration**: `npx agentdb@latest mcp`\n- **Website**: https:\u002F\u002Fagentdb.ruv.io\n\n---\n\n**Category**: Machine Learning \u002F Reinforcement Learning\n**Difficulty**: Intermediate to Advanced\n**Estimated Time**: 30-60 minutes\n",{"name":5,"description":7},"main","87afe6ce96220accf50a345dea46784c8c9f210e","YYC3-Agent-智能体\u002FYYC3-Agent-智能体-ruflo\u002Fv3\u002F@claude-flow\u002Fcli\u002F.claude\u002Fskills\u002Fagentdb-learning",0,"https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base","https:\u002F\u002Fkb.yyc3.top\u002F","https:\u002F\u002Fgithub.com\u002FYYC-Cube","万象归元于云枢 | 深栈智启新纪元","2026-11-10 10:37:32",{"source":8,"skillName":5,"commands":34,"prompt":44,"zipAvailable":45},[35,38,41],{"tool":36,"command":37},"npx","npx skills add YYC-Cube\u002FYYC3-Knowledge-Base --skill \"AgentDB Learning Plugins\"",{"tool":39,"command":40},"bunx","bunx skills add YYC-Cube\u002FYYC3-Knowledge-Base --skill \"AgentDB Learning Plugins\"",{"tool":42,"command":43},"pnpm","pnpm dlx skills add YYC-Cube\u002FYYC3-Knowledge-Base --skill \"AgentDB Learning Plugins\"","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-Agent-智能体\u002FYYC3-Agent-智能体-ruflo\u002Fv3\u002F@claude-flow\u002Fcli\u002F.claude\u002Fskills\u002Fagentdb-learning\u002FSKILL.md",true,[],[48,55,62,69,76,83],{"skillNo":49,"name":50,"nameZh":6,"description":51,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":52,"categories":53,"keywords":54,"publishedAt":19},"01M7CTDWHERCXPQCP0K03A86AS","vercel-sandbox","Run agent-browser + Chrome inside Vercel Sandbox microVMs for browser automation from any Vercel-deployed app. Use when the user needs browser automation in a Vercel app (Next.js, SvelteKit, Nuxt, Remix, Astro, etc.), wants to run headless Chrome without binary size limits, needs persistent browser sessions across commands, or wants ephemeral isolated browser environments. Triggers include \"Vercel Sandbox browser\", \"microVM Chrome\", \"agent-browser in sandbox\", \"browser automation on Vercel\", or any task requiring Chrome in a Vercel Sandbox.","https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-WebSpec-Web标准\u002FYYC3-WebSpec-Web标准-standards\u002Fbrowser-agent\u002Fskills\u002Fvercel-sandbox\u002FSKILL.md",[],[],{"skillNo":56,"name":57,"nameZh":6,"description":58,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":59,"categories":60,"keywords":61,"publishedAt":19},"01M7CTD8D2BNJY54S65M47HEJA","slack","Interact with Slack workspaces using browser automation. Use when the user needs to check unread channels, navigate Slack, send messages, extract data, find information, search conversations, or automate any Slack task. Triggers include \"check my Slack\", \"what channels have unreads\", \"send a message to\", \"search Slack for\", \"extract from Slack\", \"find who said\", or any task requiring programmatic Slack interaction.","https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-WebSpec-Web标准\u002FYYC3-WebSpec-Web标准-standards\u002Fbrowser-agent\u002Fskills\u002Fslack\u002FSKILL.md",[],[],{"skillNo":63,"name":64,"nameZh":6,"description":65,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":66,"categories":67,"keywords":68,"publishedAt":19},"01M7CTCSSB8ZP3Q58C3YZDZK1W","electron","Automate Electron desktop apps (VS Code, Slack, Discord, Figma, Notion, Spotify, etc.) using agent-browser via Chrome DevTools Protocol. Use when the user needs to interact with an Electron app, automate a desktop app, connect to a running app, control a native app, or test an Electron application. Triggers include \"automate Slack app\", \"control VS Code\", \"interact with Discord app\", \"test this Electron app\", \"connect to desktop app\", or any task requiring automation of a native Electron application.","https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-WebSpec-Web标准\u002FYYC3-WebSpec-Web标准-standards\u002Fbrowser-agent\u002Fskills\u002Felectron\u002FSKILL.md",[],[],{"skillNo":70,"name":71,"nameZh":6,"description":72,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":73,"categories":74,"keywords":75,"publishedAt":19},"01M7CTCBJY76688V4DCZGWFHGQ","dogfood","Systematically explore and test a web application to find bugs, UX issues, and other problems. Use when asked to \"dogfood\", \"QA\", \"exploratory test\", \"find issues\", \"bug hunt\", \"test this app\u002Fsite\u002Fplatform\", or review the quality of a web application. Produces a structured report with full reproduction evidence -- step-by-step screenshots, repro videos, and detailed repro steps for every issue -- so findings can be handed directly to the responsible teams.","https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-WebSpec-Web标准\u002FYYC3-WebSpec-Web标准-standards\u002Fbrowser-agent\u002Fskills\u002Fdogfood\u002FSKILL.md",[],[],{"skillNo":77,"name":78,"nameZh":6,"description":79,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":80,"categories":81,"keywords":82,"publishedAt":19},"01M7CTC4H0FFGVEAQR9PA4GA1A","agent-browser","Browser automation CLI for AI agents. Use when the user needs to interact with websites, including navigating pages, filling forms, clicking buttons, taking screenshots, extracting data, testing web apps, or automating any browser task. Triggers include requests to \"open a website\", \"fill out a form\", \"click a button\", \"take a screenshot\", \"scrape data from a page\", \"test this web app\", \"login to a site\", \"automate browser actions\", or any task requiring programmatic web interaction.","https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-WebSpec-Web标准\u002FYYC3-WebSpec-Web标准-standards\u002Fbrowser-agent\u002Fskills\u002Fagent-browser\u002FSKILL.md",[],[],{"skillNo":84,"name":85,"nameZh":6,"description":86,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":87,"categories":88,"keywords":89,"publishedAt":19},"01M7CTBKVFYBG78FDSABSDJTTQ","shadcn","Manages shadcn components and projects — adding, searching, fixing, debugging, styling, and composing UI, including chat interfaces. Provides project context, component docs, and usage examples. Applies when working with shadcn\u002Fui, component registries, presets, --preset codes, or any project with a components.json file. Also triggers for \"shadcn init\", \"create an app with --preset\", \"switch to --preset\", and for building a design system or applying a DESIGN.md or brand spec to shadcn\u002Fui.","https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-UI-界面组件\u002FYYC3-UI-界面组件-ui\u002Fskills\u002Fshadcn\u002FSKILL.md",[],[]]