[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill-letta-letta-development-guide":3,"mdc--faa5y1-key":34,"related-repo-letta-letta-development-guide":2256,"related-org-letta-letta-development-guide":2374},{"slug":4,"name":4,"fn":5,"description":6,"org":7,"tags":12,"stars":23,"repoUrl":24,"updatedAt":25,"license":26,"forks":27,"topics":28,"repo":29,"sourceUrl":32,"mdContent":33},"letta-development-guide","develop and troubleshoot Letta agents","Comprehensive guide for developing Letta agents, including architecture selection, memory design, model selection, and tool configuration. Use when building or troubleshooting Letta agents.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},"letta","Letta","https:\u002F\u002Fpexgzepcugksgbtrxkhf.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Forg-logos\u002Fletta.png","letta-ai",[13,17,20],{"name":14,"slug":15,"type":16},"Architecture","architecture","tag",{"name":18,"slug":19,"type":16},"Memory","memory",{"name":21,"slug":22,"type":16},"Agents","agents",127,"https:\u002F\u002Fgithub.com\u002Fletta-ai\u002Fskills","2026-07-13T06:25:35.232517","MIT",20,[],{"repoUrl":24,"stars":23,"forks":27,"topics":30,"description":31},[],"A shared repository for skills. Intended to be used with Letta Code, Claude Code, Codex CLI, and other agents that support skills.","https:\u002F\u002Fgithub.com\u002Fletta-ai\u002Fskills\u002Ftree\u002FHEAD\u002Fletta\u002Fagent-development","---\nname: letta-development-guide\ndescription: Comprehensive guide for developing Letta agents, including architecture selection, memory design, model selection, and tool configuration. Use when building or troubleshooting Letta agents.\nlicense: MIT\n---\n\n# Letta Development Guide\n\nComprehensive guide for designing and building effective Letta agents with appropriate architectures, memory configurations, model selection, and tool setups.\n\n## When to Use This Skill\n\nUse this skill when:\n- Starting a new Letta agent project\n- Choosing between agent architectures (letta_v1_agent vs memgpt_v2_agent)\n- Designing memory block structure and architecture\n- Selecting appropriate models for your use case\n- Planning tool configurations\n- Optimizing memory management and performance\n- Implementing shared memory between agents\n- Debugging memory-related issues\n\n## Quick Start Guide\n\n### Minimal Working Example\n\n```python\nfrom letta_client import Letta\n\nclient = Letta()\nagent = client.agents.create(\n    name=\"my-assistant\",\n    model=\"openai\u002Fgpt-4o\",\n    embedding=\"openai\u002Ftext-embedding-3-small\",\n    memory_blocks=[\n        {\"label\": \"persona\", \"value\": \"You are a helpful assistant.\"},\n        {\"label\": \"human\", \"value\": \"The user's name and preferences.\"},\n    ],\n)\n\n# Send a message\nresponse = client.agents.messages.create(\n    agent_id=agent.id,\n    messages=[{\"role\": \"user\", \"content\": \"Hello!\"}],\n)\nprint(response.messages[-1].content)\n```\n\n### 1. Architecture Selection\n\n**Use letta_v1_agent when:**\n- Building new agents (recommended default)\n- Need compatibility with reasoning models (GPT-4o, Claude Sonnet 4)\n- Want simpler system prompts and direct message generation\n\n**Use memgpt_v2_agent when:**\n- Maintaining legacy agents\n- Require specific tool patterns not yet supported in v1\n\nFor detailed comparison, see `references\u002Farchitectures.md`.\n\n### 2. Memory Architecture Design\n\nMemory is the foundation of effective agents. Letta provides three memory types:\n\n**Core Memory (in-context):**\n- Always accessible in agent's context window\n- Use for: current state, active context, frequently referenced information\n- Limit: Keep total core memory under 80% of context window\n\n**Archival Memory (out-of-context):**\n- Semantic search over vector database\n- Use for: historical records, large knowledge bases, past interactions\n- Access: Agent must explicitly call archival_memory_search\n- Note: NOT automatically populated from context overflow\n\n**Conversation History:**\n- Past messages from current conversation\n- Retrieved via conversation_search tool\n- Use for: referencing earlier discussion, tracking conversation flow\n\nSee `references\u002Fmemory-architecture.md` for detailed guidance.\n\n### 3. Memory Block Design\n\n**Core principle:** One block per distinct functional unit.\n\n**Essential blocks:**\n- `persona`: Agent identity, behavioral guidelines, capabilities\n- `human`: User information, preferences, context\n\n**Add domain-specific blocks based on use case:**\n- Customer support: `company_policies`, `product_knowledge`, `customer`\n- Coding assistant: `project_context`, `coding_standards`, `current_task`\n- Personal assistant: `schedule`, `preferences`, `contacts`\n\n**Memory block guidelines:**\n- Keep blocks focused and purpose-specific\n- Use clear, instructional descriptions\n- Monitor size limits (typically 2000-5000 characters per block)\n- Design for append operations when sharing memory between agents\n\nSee `references\u002Fmemory-patterns.md` for domain examples and `references\u002Fdescription-patterns.md` for writing effective descriptions.\n\n### 4. Model Selection\n\nMatch model capabilities to agent requirements:\n\n**For production agents:**\n- GPT-4o or Claude Sonnet 4 for complex reasoning\n- GPT-4o-mini for cost-efficient general tasks\n- Claude Haiku 3.5 for fast, lightweight operations\n- Gemini 2.0 Flash for balanced speed\u002Fcapability\n\n**Avoid for production:**\n- Small Ollama models (\u003C7B parameters) - poor tool calling\n- Models without reliable function calling support\n\nSee `references\u002Fmodel-recommendations.md` for detailed guidance.\n\n### 5. Tool Configuration\n\n**Start minimal:** Attach only tools the agent will actively use.\n\n**Common starting points:**\n- **Memory tools** (memory_insert, memory_replace, memory_rethink): Core for most agents\n- **File system tools**: Auto-attached when folders are connected\n- **Custom tools**: For domain-specific operations (databases, APIs, etc.)\n\n**Tool Rules:** Use to enforce sequencing when needed (e.g., \"always call search before answer\")\n\nConsult `references\u002Ftool-patterns.md` for common configurations.\n\n## Advanced Topics\n\n### Memory Size Management\n\n**When approaching character limits:**\n1. **Split by topic:** `customer_profile` → `customer_business`, `customer_preferences`\n2. **Split by time:** `interaction_history` → `recent_interactions`, archive older to archival memory\n3. **Archive historical data:** Move old information to archival memory\n4. **Consolidate with memory_rethink:** Summarize and rewrite block\n\nSee `references\u002Fsize-management.md` for strategies.\n\n### Concurrency Patterns\n\nWhen multiple agents share memory blocks or an agent processes concurrent requests:\n\n**Safest operations:**\n- `memory_insert`: Append-only, minimal race conditions\n- Database uses PostgreSQL row-level locking\n\n**Risk of race conditions:**\n- `memory_replace`: Target string may change before write\n- `memory_rethink`: Last-writer-wins, no merge\n\n**Best practices:**\n- Design for append operations when possible\n- Use memory_insert for concurrent writes\n- Reserve memory_rethink for single-agent exclusive access\n\nConsult `references\u002Fconcurrency.md` for detailed patterns.\n\n## Validation Checklist\n\nBefore finalizing your agent design:\n\n**Architecture:**\n- [ ] Does the architecture match the model's capabilities?\n- [ ] Is the model appropriate for expected workload and latency requirements?\n\n**Memory:**\n- [ ] Is core memory total under 80% of context window?\n- [ ] Is each block focused on one functional area?\n- [ ] Are descriptions clear about when to read\u002Fwrite?\n- [ ] Have you planned for size growth and overflow?\n- [ ] If multi-agent, are concurrency patterns considered?\n\n**Tools:**\n- [ ] Are tools necessary and properly configured?\n- [ ] Are memory blocks granular enough for effective updates?\n\n## Common Antipatterns\n\n**Too few memory blocks:**\n```yaml\n# Bad: Everything in one block\nagent_memory: \"Agent is helpful. User is John...\"\n```\nSplit into focused blocks instead.\n\n**Too many memory blocks:**\nCreating 10+ blocks when 3-4 would suffice. Start minimal, expand as needed.\n\n**Poor descriptions:**\n```yaml\n# Bad\ndata: \"Contains data\"\n```\nProvide actionable guidance instead. See `references\u002Fdescription-patterns.md`.\n\n**Ignoring size limits:**\nLetting blocks grow indefinitely until they hit limits. Monitor and manage proactively.\n\n## Implementation Steps\n\n### 1. Design Phase\n- Choose architecture based on requirements\n- Design memory block structure\n- Select appropriate model\n- Plan tool configuration\n\n### 2. Creation Phase (SDK)\n\n**Python:**\n```python\nfrom letta_client import Letta\n\nclient = Letta()  # Uses LETTA_API_KEY env var\n\n# Create agent with custom memory blocks\nagent = client.agents.create(\n    name=\"my-agent\",\n    model=\"openai\u002Fgpt-4o\",  # or \"anthropic\u002Fclaude-sonnet-4-20250514\"\n    embedding=\"openai\u002Ftext-embedding-3-small\",\n    memory_blocks=[\n        {\"label\": \"persona\", \"value\": \"You are a helpful assistant...\"},\n        {\"label\": \"human\", \"value\": \"User preferences and context...\"},\n        {\"label\": \"project\", \"value\": \"Current project details...\"},\n    ],\n    description=\"Agent for helping with X\",\n)\nprint(f\"Created agent: {agent.id}\")\n```\n\n**TypeScript:**\n```typescript\nimport Letta from \"letta-client\";\n\nconst client = new Letta();\n\nconst agent = await client.agents.create({\n  name: \"my-agent\",\n  model: \"openai\u002Fgpt-4o\",\n  embedding: \"openai\u002Ftext-embedding-3-small\",\n  memoryBlocks: [\n    { label: \"persona\", value: \"You are a helpful assistant...\" },\n    { label: \"human\", value: \"User preferences and context...\" },\n    { label: \"project\", value: \"Current project details...\" },\n  ],\n  description: \"Agent for helping with X\",\n});\nconsole.log(`Created agent: ${agent.id}`);\n```\n\n**Note:** Letta Code CLI (`letta` command) creates agents interactively. Use `letta --new-agent` to start fresh, then `\u002Frename` and `\u002Fdescription` to configure.\n\n### 3. Testing Phase\n- Test with representative queries\n- Monitor memory tool usage patterns\n- Verify tool calling behavior\n\n### 4. Iteration Phase\n- Refine memory block structure based on actual usage\n- Optimize system instructions\n- Adjust tool configurations\n\n## References\n\nFor detailed information on specific topics, consult the reference materials:\n\n- `references\u002Farchitectures.md` - Architecture comparison and selection\n- `references\u002Fmemory-architecture.md` - Memory types and when to use them\n- `references\u002Fmemory-patterns.md` - Domain-specific memory block examples\n- `references\u002Fdescription-patterns.md` - Writing effective block descriptions\n- `references\u002Fsize-management.md` - Managing memory block size limits\n- `references\u002Fconcurrency.md` - Multi-agent memory sharing patterns\n- `references\u002Fmodel-recommendations.md` - Model selection guidance\n- `references\u002Ftool-patterns.md` - Common tool configurations\n",{"data":35,"body":36},{"name":4,"description":6,"license":26},{"type":37,"children":38},"root",[39,47,53,60,65,110,116,123,307,313,322,340,348,361,374,380,385,393,411,419,442,450,468,481,487,497,505,530,538,617,625,648,668,674,679,687,710,718,731,742,748,758,766,799,809,822,828,834,842,921,933,939,944,952,971,979,1004,1012,1030,1042,1048,1053,1061,1087,1095,1144,1152,1174,1180,1188,1237,1242,1252,1260,1300,1311,1321,1327,1333,1356,1362,1370,1505,1513,2067,2108,2114,2132,2138,2156,2162,2167,2250],{"type":40,"tag":41,"props":42,"children":43},"element","h1",{"id":4},[44],{"type":45,"value":46},"text","Letta Development Guide",{"type":40,"tag":48,"props":49,"children":50},"p",{},[51],{"type":45,"value":52},"Comprehensive guide for designing and building effective Letta agents with appropriate architectures, memory configurations, model selection, and tool setups.",{"type":40,"tag":54,"props":55,"children":57},"h2",{"id":56},"when-to-use-this-skill",[58],{"type":45,"value":59},"When to Use This Skill",{"type":40,"tag":48,"props":61,"children":62},{},[63],{"type":45,"value":64},"Use this skill when:",{"type":40,"tag":66,"props":67,"children":68},"ul",{},[69,75,80,85,90,95,100,105],{"type":40,"tag":70,"props":71,"children":72},"li",{},[73],{"type":45,"value":74},"Starting a new Letta agent project",{"type":40,"tag":70,"props":76,"children":77},{},[78],{"type":45,"value":79},"Choosing between agent architectures (letta_v1_agent vs memgpt_v2_agent)",{"type":40,"tag":70,"props":81,"children":82},{},[83],{"type":45,"value":84},"Designing memory block structure and architecture",{"type":40,"tag":70,"props":86,"children":87},{},[88],{"type":45,"value":89},"Selecting appropriate models for your use case",{"type":40,"tag":70,"props":91,"children":92},{},[93],{"type":45,"value":94},"Planning tool configurations",{"type":40,"tag":70,"props":96,"children":97},{},[98],{"type":45,"value":99},"Optimizing memory management and performance",{"type":40,"tag":70,"props":101,"children":102},{},[103],{"type":45,"value":104},"Implementing shared memory between agents",{"type":40,"tag":70,"props":106,"children":107},{},[108],{"type":45,"value":109},"Debugging memory-related issues",{"type":40,"tag":54,"props":111,"children":113},{"id":112},"quick-start-guide",[114],{"type":45,"value":115},"Quick Start Guide",{"type":40,"tag":117,"props":118,"children":120},"h3",{"id":119},"minimal-working-example",[121],{"type":45,"value":122},"Minimal Working Example",{"type":40,"tag":124,"props":125,"children":130},"pre",{"className":126,"code":127,"language":128,"meta":129,"style":129},"language-python shiki shiki-themes material-theme-lighter material-theme material-theme-palenight","from letta_client import Letta\n\nclient = Letta()\nagent = client.agents.create(\n    name=\"my-assistant\",\n    model=\"openai\u002Fgpt-4o\",\n    embedding=\"openai\u002Ftext-embedding-3-small\",\n    memory_blocks=[\n        {\"label\": \"persona\", \"value\": \"You are a helpful assistant.\"},\n        {\"label\": \"human\", \"value\": \"The user's name and preferences.\"},\n    ],\n)\n\n# Send a message\nresponse = client.agents.messages.create(\n    agent_id=agent.id,\n    messages=[{\"role\": \"user\", \"content\": \"Hello!\"}],\n)\nprint(response.messages[-1].content)\n","python","",[131],{"type":40,"tag":132,"props":133,"children":134},"code",{"__ignoreMap":129},[135,146,156,165,174,183,192,201,210,219,228,237,246,254,263,272,281,290,298],{"type":40,"tag":136,"props":137,"children":140},"span",{"class":138,"line":139},"line",1,[141],{"type":40,"tag":136,"props":142,"children":143},{},[144],{"type":45,"value":145},"from letta_client import Letta\n",{"type":40,"tag":136,"props":147,"children":149},{"class":138,"line":148},2,[150],{"type":40,"tag":136,"props":151,"children":153},{"emptyLinePlaceholder":152},true,[154],{"type":45,"value":155},"\n",{"type":40,"tag":136,"props":157,"children":159},{"class":138,"line":158},3,[160],{"type":40,"tag":136,"props":161,"children":162},{},[163],{"type":45,"value":164},"client = Letta()\n",{"type":40,"tag":136,"props":166,"children":168},{"class":138,"line":167},4,[169],{"type":40,"tag":136,"props":170,"children":171},{},[172],{"type":45,"value":173},"agent = client.agents.create(\n",{"type":40,"tag":136,"props":175,"children":177},{"class":138,"line":176},5,[178],{"type":40,"tag":136,"props":179,"children":180},{},[181],{"type":45,"value":182},"    name=\"my-assistant\",\n",{"type":40,"tag":136,"props":184,"children":186},{"class":138,"line":185},6,[187],{"type":40,"tag":136,"props":188,"children":189},{},[190],{"type":45,"value":191},"    model=\"openai\u002Fgpt-4o\",\n",{"type":40,"tag":136,"props":193,"children":195},{"class":138,"line":194},7,[196],{"type":40,"tag":136,"props":197,"children":198},{},[199],{"type":45,"value":200},"    embedding=\"openai\u002Ftext-embedding-3-small\",\n",{"type":40,"tag":136,"props":202,"children":204},{"class":138,"line":203},8,[205],{"type":40,"tag":136,"props":206,"children":207},{},[208],{"type":45,"value":209},"    memory_blocks=[\n",{"type":40,"tag":136,"props":211,"children":213},{"class":138,"line":212},9,[214],{"type":40,"tag":136,"props":215,"children":216},{},[217],{"type":45,"value":218},"        {\"label\": \"persona\", \"value\": \"You are a helpful assistant.\"},\n",{"type":40,"tag":136,"props":220,"children":222},{"class":138,"line":221},10,[223],{"type":40,"tag":136,"props":224,"children":225},{},[226],{"type":45,"value":227},"        {\"label\": \"human\", \"value\": \"The user's name and preferences.\"},\n",{"type":40,"tag":136,"props":229,"children":231},{"class":138,"line":230},11,[232],{"type":40,"tag":136,"props":233,"children":234},{},[235],{"type":45,"value":236},"    ],\n",{"type":40,"tag":136,"props":238,"children":240},{"class":138,"line":239},12,[241],{"type":40,"tag":136,"props":242,"children":243},{},[244],{"type":45,"value":245},")\n",{"type":40,"tag":136,"props":247,"children":249},{"class":138,"line":248},13,[250],{"type":40,"tag":136,"props":251,"children":252},{"emptyLinePlaceholder":152},[253],{"type":45,"value":155},{"type":40,"tag":136,"props":255,"children":257},{"class":138,"line":256},14,[258],{"type":40,"tag":136,"props":259,"children":260},{},[261],{"type":45,"value":262},"# Send a message\n",{"type":40,"tag":136,"props":264,"children":266},{"class":138,"line":265},15,[267],{"type":40,"tag":136,"props":268,"children":269},{},[270],{"type":45,"value":271},"response = client.agents.messages.create(\n",{"type":40,"tag":136,"props":273,"children":275},{"class":138,"line":274},16,[276],{"type":40,"tag":136,"props":277,"children":278},{},[279],{"type":45,"value":280},"    agent_id=agent.id,\n",{"type":40,"tag":136,"props":282,"children":284},{"class":138,"line":283},17,[285],{"type":40,"tag":136,"props":286,"children":287},{},[288],{"type":45,"value":289},"    messages=[{\"role\": \"user\", \"content\": \"Hello!\"}],\n",{"type":40,"tag":136,"props":291,"children":293},{"class":138,"line":292},18,[294],{"type":40,"tag":136,"props":295,"children":296},{},[297],{"type":45,"value":245},{"type":40,"tag":136,"props":299,"children":301},{"class":138,"line":300},19,[302],{"type":40,"tag":136,"props":303,"children":304},{},[305],{"type":45,"value":306},"print(response.messages[-1].content)\n",{"type":40,"tag":117,"props":308,"children":310},{"id":309},"_1-architecture-selection",[311],{"type":45,"value":312},"1. Architecture Selection",{"type":40,"tag":48,"props":314,"children":315},{},[316],{"type":40,"tag":317,"props":318,"children":319},"strong",{},[320],{"type":45,"value":321},"Use letta_v1_agent when:",{"type":40,"tag":66,"props":323,"children":324},{},[325,330,335],{"type":40,"tag":70,"props":326,"children":327},{},[328],{"type":45,"value":329},"Building new agents (recommended default)",{"type":40,"tag":70,"props":331,"children":332},{},[333],{"type":45,"value":334},"Need compatibility with reasoning models (GPT-4o, Claude Sonnet 4)",{"type":40,"tag":70,"props":336,"children":337},{},[338],{"type":45,"value":339},"Want simpler system prompts and direct message generation",{"type":40,"tag":48,"props":341,"children":342},{},[343],{"type":40,"tag":317,"props":344,"children":345},{},[346],{"type":45,"value":347},"Use memgpt_v2_agent when:",{"type":40,"tag":66,"props":349,"children":350},{},[351,356],{"type":40,"tag":70,"props":352,"children":353},{},[354],{"type":45,"value":355},"Maintaining legacy agents",{"type":40,"tag":70,"props":357,"children":358},{},[359],{"type":45,"value":360},"Require specific tool patterns not yet supported in v1",{"type":40,"tag":48,"props":362,"children":363},{},[364,366,372],{"type":45,"value":365},"For detailed comparison, see ",{"type":40,"tag":132,"props":367,"children":369},{"className":368},[],[370],{"type":45,"value":371},"references\u002Farchitectures.md",{"type":45,"value":373},".",{"type":40,"tag":117,"props":375,"children":377},{"id":376},"_2-memory-architecture-design",[378],{"type":45,"value":379},"2. Memory Architecture Design",{"type":40,"tag":48,"props":381,"children":382},{},[383],{"type":45,"value":384},"Memory is the foundation of effective agents. Letta provides three memory types:",{"type":40,"tag":48,"props":386,"children":387},{},[388],{"type":40,"tag":317,"props":389,"children":390},{},[391],{"type":45,"value":392},"Core Memory (in-context):",{"type":40,"tag":66,"props":394,"children":395},{},[396,401,406],{"type":40,"tag":70,"props":397,"children":398},{},[399],{"type":45,"value":400},"Always accessible in agent's context window",{"type":40,"tag":70,"props":402,"children":403},{},[404],{"type":45,"value":405},"Use for: current state, active context, frequently referenced information",{"type":40,"tag":70,"props":407,"children":408},{},[409],{"type":45,"value":410},"Limit: Keep total core memory under 80% of context window",{"type":40,"tag":48,"props":412,"children":413},{},[414],{"type":40,"tag":317,"props":415,"children":416},{},[417],{"type":45,"value":418},"Archival Memory (out-of-context):",{"type":40,"tag":66,"props":420,"children":421},{},[422,427,432,437],{"type":40,"tag":70,"props":423,"children":424},{},[425],{"type":45,"value":426},"Semantic search over vector database",{"type":40,"tag":70,"props":428,"children":429},{},[430],{"type":45,"value":431},"Use for: historical records, large knowledge bases, past interactions",{"type":40,"tag":70,"props":433,"children":434},{},[435],{"type":45,"value":436},"Access: Agent must explicitly call archival_memory_search",{"type":40,"tag":70,"props":438,"children":439},{},[440],{"type":45,"value":441},"Note: NOT automatically populated from context overflow",{"type":40,"tag":48,"props":443,"children":444},{},[445],{"type":40,"tag":317,"props":446,"children":447},{},[448],{"type":45,"value":449},"Conversation History:",{"type":40,"tag":66,"props":451,"children":452},{},[453,458,463],{"type":40,"tag":70,"props":454,"children":455},{},[456],{"type":45,"value":457},"Past messages from current conversation",{"type":40,"tag":70,"props":459,"children":460},{},[461],{"type":45,"value":462},"Retrieved via conversation_search tool",{"type":40,"tag":70,"props":464,"children":465},{},[466],{"type":45,"value":467},"Use for: referencing earlier discussion, tracking conversation flow",{"type":40,"tag":48,"props":469,"children":470},{},[471,473,479],{"type":45,"value":472},"See ",{"type":40,"tag":132,"props":474,"children":476},{"className":475},[],[477],{"type":45,"value":478},"references\u002Fmemory-architecture.md",{"type":45,"value":480}," for detailed guidance.",{"type":40,"tag":117,"props":482,"children":484},{"id":483},"_3-memory-block-design",[485],{"type":45,"value":486},"3. Memory Block Design",{"type":40,"tag":48,"props":488,"children":489},{},[490,495],{"type":40,"tag":317,"props":491,"children":492},{},[493],{"type":45,"value":494},"Core principle:",{"type":45,"value":496}," One block per distinct functional unit.",{"type":40,"tag":48,"props":498,"children":499},{},[500],{"type":40,"tag":317,"props":501,"children":502},{},[503],{"type":45,"value":504},"Essential blocks:",{"type":40,"tag":66,"props":506,"children":507},{},[508,519],{"type":40,"tag":70,"props":509,"children":510},{},[511,517],{"type":40,"tag":132,"props":512,"children":514},{"className":513},[],[515],{"type":45,"value":516},"persona",{"type":45,"value":518},": Agent identity, behavioral guidelines, capabilities",{"type":40,"tag":70,"props":520,"children":521},{},[522,528],{"type":40,"tag":132,"props":523,"children":525},{"className":524},[],[526],{"type":45,"value":527},"human",{"type":45,"value":529},": User information, preferences, context",{"type":40,"tag":48,"props":531,"children":532},{},[533],{"type":40,"tag":317,"props":534,"children":535},{},[536],{"type":45,"value":537},"Add domain-specific blocks based on use case:",{"type":40,"tag":66,"props":539,"children":540},{},[541,567,592],{"type":40,"tag":70,"props":542,"children":543},{},[544,546,552,554,560,561],{"type":45,"value":545},"Customer support: ",{"type":40,"tag":132,"props":547,"children":549},{"className":548},[],[550],{"type":45,"value":551},"company_policies",{"type":45,"value":553},", ",{"type":40,"tag":132,"props":555,"children":557},{"className":556},[],[558],{"type":45,"value":559},"product_knowledge",{"type":45,"value":553},{"type":40,"tag":132,"props":562,"children":564},{"className":563},[],[565],{"type":45,"value":566},"customer",{"type":40,"tag":70,"props":568,"children":569},{},[570,572,578,579,585,586],{"type":45,"value":571},"Coding assistant: ",{"type":40,"tag":132,"props":573,"children":575},{"className":574},[],[576],{"type":45,"value":577},"project_context",{"type":45,"value":553},{"type":40,"tag":132,"props":580,"children":582},{"className":581},[],[583],{"type":45,"value":584},"coding_standards",{"type":45,"value":553},{"type":40,"tag":132,"props":587,"children":589},{"className":588},[],[590],{"type":45,"value":591},"current_task",{"type":40,"tag":70,"props":593,"children":594},{},[595,597,603,604,610,611],{"type":45,"value":596},"Personal assistant: ",{"type":40,"tag":132,"props":598,"children":600},{"className":599},[],[601],{"type":45,"value":602},"schedule",{"type":45,"value":553},{"type":40,"tag":132,"props":605,"children":607},{"className":606},[],[608],{"type":45,"value":609},"preferences",{"type":45,"value":553},{"type":40,"tag":132,"props":612,"children":614},{"className":613},[],[615],{"type":45,"value":616},"contacts",{"type":40,"tag":48,"props":618,"children":619},{},[620],{"type":40,"tag":317,"props":621,"children":622},{},[623],{"type":45,"value":624},"Memory block guidelines:",{"type":40,"tag":66,"props":626,"children":627},{},[628,633,638,643],{"type":40,"tag":70,"props":629,"children":630},{},[631],{"type":45,"value":632},"Keep blocks focused and purpose-specific",{"type":40,"tag":70,"props":634,"children":635},{},[636],{"type":45,"value":637},"Use clear, instructional descriptions",{"type":40,"tag":70,"props":639,"children":640},{},[641],{"type":45,"value":642},"Monitor size limits (typically 2000-5000 characters per block)",{"type":40,"tag":70,"props":644,"children":645},{},[646],{"type":45,"value":647},"Design for append operations when sharing memory between agents",{"type":40,"tag":48,"props":649,"children":650},{},[651,652,658,660,666],{"type":45,"value":472},{"type":40,"tag":132,"props":653,"children":655},{"className":654},[],[656],{"type":45,"value":657},"references\u002Fmemory-patterns.md",{"type":45,"value":659}," for domain examples and ",{"type":40,"tag":132,"props":661,"children":663},{"className":662},[],[664],{"type":45,"value":665},"references\u002Fdescription-patterns.md",{"type":45,"value":667}," for writing effective descriptions.",{"type":40,"tag":117,"props":669,"children":671},{"id":670},"_4-model-selection",[672],{"type":45,"value":673},"4. Model Selection",{"type":40,"tag":48,"props":675,"children":676},{},[677],{"type":45,"value":678},"Match model capabilities to agent requirements:",{"type":40,"tag":48,"props":680,"children":681},{},[682],{"type":40,"tag":317,"props":683,"children":684},{},[685],{"type":45,"value":686},"For production agents:",{"type":40,"tag":66,"props":688,"children":689},{},[690,695,700,705],{"type":40,"tag":70,"props":691,"children":692},{},[693],{"type":45,"value":694},"GPT-4o or Claude Sonnet 4 for complex reasoning",{"type":40,"tag":70,"props":696,"children":697},{},[698],{"type":45,"value":699},"GPT-4o-mini for cost-efficient general tasks",{"type":40,"tag":70,"props":701,"children":702},{},[703],{"type":45,"value":704},"Claude Haiku 3.5 for fast, lightweight operations",{"type":40,"tag":70,"props":706,"children":707},{},[708],{"type":45,"value":709},"Gemini 2.0 Flash for balanced speed\u002Fcapability",{"type":40,"tag":48,"props":711,"children":712},{},[713],{"type":40,"tag":317,"props":714,"children":715},{},[716],{"type":45,"value":717},"Avoid for production:",{"type":40,"tag":66,"props":719,"children":720},{},[721,726],{"type":40,"tag":70,"props":722,"children":723},{},[724],{"type":45,"value":725},"Small Ollama models (\u003C7B parameters) - poor tool calling",{"type":40,"tag":70,"props":727,"children":728},{},[729],{"type":45,"value":730},"Models without reliable function calling support",{"type":40,"tag":48,"props":732,"children":733},{},[734,735,741],{"type":45,"value":472},{"type":40,"tag":132,"props":736,"children":738},{"className":737},[],[739],{"type":45,"value":740},"references\u002Fmodel-recommendations.md",{"type":45,"value":480},{"type":40,"tag":117,"props":743,"children":745},{"id":744},"_5-tool-configuration",[746],{"type":45,"value":747},"5. Tool Configuration",{"type":40,"tag":48,"props":749,"children":750},{},[751,756],{"type":40,"tag":317,"props":752,"children":753},{},[754],{"type":45,"value":755},"Start minimal:",{"type":45,"value":757}," Attach only tools the agent will actively use.",{"type":40,"tag":48,"props":759,"children":760},{},[761],{"type":40,"tag":317,"props":762,"children":763},{},[764],{"type":45,"value":765},"Common starting points:",{"type":40,"tag":66,"props":767,"children":768},{},[769,779,789],{"type":40,"tag":70,"props":770,"children":771},{},[772,777],{"type":40,"tag":317,"props":773,"children":774},{},[775],{"type":45,"value":776},"Memory tools",{"type":45,"value":778}," (memory_insert, memory_replace, memory_rethink): Core for most agents",{"type":40,"tag":70,"props":780,"children":781},{},[782,787],{"type":40,"tag":317,"props":783,"children":784},{},[785],{"type":45,"value":786},"File system tools",{"type":45,"value":788},": Auto-attached when folders are connected",{"type":40,"tag":70,"props":790,"children":791},{},[792,797],{"type":40,"tag":317,"props":793,"children":794},{},[795],{"type":45,"value":796},"Custom tools",{"type":45,"value":798},": For domain-specific operations (databases, APIs, etc.)",{"type":40,"tag":48,"props":800,"children":801},{},[802,807],{"type":40,"tag":317,"props":803,"children":804},{},[805],{"type":45,"value":806},"Tool Rules:",{"type":45,"value":808}," Use to enforce sequencing when needed (e.g., \"always call search before answer\")",{"type":40,"tag":48,"props":810,"children":811},{},[812,814,820],{"type":45,"value":813},"Consult ",{"type":40,"tag":132,"props":815,"children":817},{"className":816},[],[818],{"type":45,"value":819},"references\u002Ftool-patterns.md",{"type":45,"value":821}," for common configurations.",{"type":40,"tag":54,"props":823,"children":825},{"id":824},"advanced-topics",[826],{"type":45,"value":827},"Advanced Topics",{"type":40,"tag":117,"props":829,"children":831},{"id":830},"memory-size-management",[832],{"type":45,"value":833},"Memory Size Management",{"type":40,"tag":48,"props":835,"children":836},{},[837],{"type":40,"tag":317,"props":838,"children":839},{},[840],{"type":45,"value":841},"When approaching character limits:",{"type":40,"tag":843,"props":844,"children":845},"ol",{},[846,877,901,911],{"type":40,"tag":70,"props":847,"children":848},{},[849,854,856,862,864,870,871],{"type":40,"tag":317,"props":850,"children":851},{},[852],{"type":45,"value":853},"Split by topic:",{"type":45,"value":855}," ",{"type":40,"tag":132,"props":857,"children":859},{"className":858},[],[860],{"type":45,"value":861},"customer_profile",{"type":45,"value":863}," → ",{"type":40,"tag":132,"props":865,"children":867},{"className":866},[],[868],{"type":45,"value":869},"customer_business",{"type":45,"value":553},{"type":40,"tag":132,"props":872,"children":874},{"className":873},[],[875],{"type":45,"value":876},"customer_preferences",{"type":40,"tag":70,"props":878,"children":879},{},[880,885,886,892,893,899],{"type":40,"tag":317,"props":881,"children":882},{},[883],{"type":45,"value":884},"Split by time:",{"type":45,"value":855},{"type":40,"tag":132,"props":887,"children":889},{"className":888},[],[890],{"type":45,"value":891},"interaction_history",{"type":45,"value":863},{"type":40,"tag":132,"props":894,"children":896},{"className":895},[],[897],{"type":45,"value":898},"recent_interactions",{"type":45,"value":900},", archive older to archival memory",{"type":40,"tag":70,"props":902,"children":903},{},[904,909],{"type":40,"tag":317,"props":905,"children":906},{},[907],{"type":45,"value":908},"Archive historical data:",{"type":45,"value":910}," Move old information to archival memory",{"type":40,"tag":70,"props":912,"children":913},{},[914,919],{"type":40,"tag":317,"props":915,"children":916},{},[917],{"type":45,"value":918},"Consolidate with memory_rethink:",{"type":45,"value":920}," Summarize and rewrite block",{"type":40,"tag":48,"props":922,"children":923},{},[924,925,931],{"type":45,"value":472},{"type":40,"tag":132,"props":926,"children":928},{"className":927},[],[929],{"type":45,"value":930},"references\u002Fsize-management.md",{"type":45,"value":932}," for strategies.",{"type":40,"tag":117,"props":934,"children":936},{"id":935},"concurrency-patterns",[937],{"type":45,"value":938},"Concurrency Patterns",{"type":40,"tag":48,"props":940,"children":941},{},[942],{"type":45,"value":943},"When multiple agents share memory blocks or an agent processes concurrent requests:",{"type":40,"tag":48,"props":945,"children":946},{},[947],{"type":40,"tag":317,"props":948,"children":949},{},[950],{"type":45,"value":951},"Safest operations:",{"type":40,"tag":66,"props":953,"children":954},{},[955,966],{"type":40,"tag":70,"props":956,"children":957},{},[958,964],{"type":40,"tag":132,"props":959,"children":961},{"className":960},[],[962],{"type":45,"value":963},"memory_insert",{"type":45,"value":965},": Append-only, minimal race conditions",{"type":40,"tag":70,"props":967,"children":968},{},[969],{"type":45,"value":970},"Database uses PostgreSQL row-level locking",{"type":40,"tag":48,"props":972,"children":973},{},[974],{"type":40,"tag":317,"props":975,"children":976},{},[977],{"type":45,"value":978},"Risk of race conditions:",{"type":40,"tag":66,"props":980,"children":981},{},[982,993],{"type":40,"tag":70,"props":983,"children":984},{},[985,991],{"type":40,"tag":132,"props":986,"children":988},{"className":987},[],[989],{"type":45,"value":990},"memory_replace",{"type":45,"value":992},": Target string may change before write",{"type":40,"tag":70,"props":994,"children":995},{},[996,1002],{"type":40,"tag":132,"props":997,"children":999},{"className":998},[],[1000],{"type":45,"value":1001},"memory_rethink",{"type":45,"value":1003},": Last-writer-wins, no merge",{"type":40,"tag":48,"props":1005,"children":1006},{},[1007],{"type":40,"tag":317,"props":1008,"children":1009},{},[1010],{"type":45,"value":1011},"Best practices:",{"type":40,"tag":66,"props":1013,"children":1014},{},[1015,1020,1025],{"type":40,"tag":70,"props":1016,"children":1017},{},[1018],{"type":45,"value":1019},"Design for append operations when possible",{"type":40,"tag":70,"props":1021,"children":1022},{},[1023],{"type":45,"value":1024},"Use memory_insert for concurrent writes",{"type":40,"tag":70,"props":1026,"children":1027},{},[1028],{"type":45,"value":1029},"Reserve memory_rethink for single-agent exclusive access",{"type":40,"tag":48,"props":1031,"children":1032},{},[1033,1034,1040],{"type":45,"value":813},{"type":40,"tag":132,"props":1035,"children":1037},{"className":1036},[],[1038],{"type":45,"value":1039},"references\u002Fconcurrency.md",{"type":45,"value":1041}," for detailed patterns.",{"type":40,"tag":54,"props":1043,"children":1045},{"id":1044},"validation-checklist",[1046],{"type":45,"value":1047},"Validation Checklist",{"type":40,"tag":48,"props":1049,"children":1050},{},[1051],{"type":45,"value":1052},"Before finalizing your agent design:",{"type":40,"tag":48,"props":1054,"children":1055},{},[1056],{"type":40,"tag":317,"props":1057,"children":1058},{},[1059],{"type":45,"value":1060},"Architecture:",{"type":40,"tag":66,"props":1062,"children":1065},{"className":1063},[1064],"contains-task-list",[1066,1078],{"type":40,"tag":70,"props":1067,"children":1070},{"className":1068},[1069],"task-list-item",[1071,1076],{"type":40,"tag":1072,"props":1073,"children":1075},"input",{"disabled":152,"type":1074},"checkbox",[],{"type":45,"value":1077}," Does the architecture match the model's capabilities?",{"type":40,"tag":70,"props":1079,"children":1081},{"className":1080},[1069],[1082,1085],{"type":40,"tag":1072,"props":1083,"children":1084},{"disabled":152,"type":1074},[],{"type":45,"value":1086}," Is the model appropriate for expected workload and latency requirements?",{"type":40,"tag":48,"props":1088,"children":1089},{},[1090],{"type":40,"tag":317,"props":1091,"children":1092},{},[1093],{"type":45,"value":1094},"Memory:",{"type":40,"tag":66,"props":1096,"children":1098},{"className":1097},[1064],[1099,1108,1117,1126,1135],{"type":40,"tag":70,"props":1100,"children":1102},{"className":1101},[1069],[1103,1106],{"type":40,"tag":1072,"props":1104,"children":1105},{"disabled":152,"type":1074},[],{"type":45,"value":1107}," Is core memory total under 80% of context window?",{"type":40,"tag":70,"props":1109,"children":1111},{"className":1110},[1069],[1112,1115],{"type":40,"tag":1072,"props":1113,"children":1114},{"disabled":152,"type":1074},[],{"type":45,"value":1116}," Is each block focused on one functional area?",{"type":40,"tag":70,"props":1118,"children":1120},{"className":1119},[1069],[1121,1124],{"type":40,"tag":1072,"props":1122,"children":1123},{"disabled":152,"type":1074},[],{"type":45,"value":1125}," Are descriptions clear about when to read\u002Fwrite?",{"type":40,"tag":70,"props":1127,"children":1129},{"className":1128},[1069],[1130,1133],{"type":40,"tag":1072,"props":1131,"children":1132},{"disabled":152,"type":1074},[],{"type":45,"value":1134}," Have you planned for size growth and overflow?",{"type":40,"tag":70,"props":1136,"children":1138},{"className":1137},[1069],[1139,1142],{"type":40,"tag":1072,"props":1140,"children":1141},{"disabled":152,"type":1074},[],{"type":45,"value":1143}," If multi-agent, are concurrency patterns considered?",{"type":40,"tag":48,"props":1145,"children":1146},{},[1147],{"type":40,"tag":317,"props":1148,"children":1149},{},[1150],{"type":45,"value":1151},"Tools:",{"type":40,"tag":66,"props":1153,"children":1155},{"className":1154},[1064],[1156,1165],{"type":40,"tag":70,"props":1157,"children":1159},{"className":1158},[1069],[1160,1163],{"type":40,"tag":1072,"props":1161,"children":1162},{"disabled":152,"type":1074},[],{"type":45,"value":1164}," Are tools necessary and properly configured?",{"type":40,"tag":70,"props":1166,"children":1168},{"className":1167},[1069],[1169,1172],{"type":40,"tag":1072,"props":1170,"children":1171},{"disabled":152,"type":1074},[],{"type":45,"value":1173}," Are memory blocks granular enough for effective updates?",{"type":40,"tag":54,"props":1175,"children":1177},{"id":1176},"common-antipatterns",[1178],{"type":45,"value":1179},"Common Antipatterns",{"type":40,"tag":48,"props":1181,"children":1182},{},[1183],{"type":40,"tag":317,"props":1184,"children":1185},{},[1186],{"type":45,"value":1187},"Too few memory blocks:",{"type":40,"tag":124,"props":1189,"children":1193},{"className":1190,"code":1191,"language":1192,"meta":129,"style":129},"language-yaml shiki shiki-themes material-theme-lighter material-theme material-theme-palenight","# Bad: Everything in one block\nagent_memory: \"Agent is helpful. User is John...\"\n","yaml",[1194],{"type":40,"tag":132,"props":1195,"children":1196},{"__ignoreMap":129},[1197,1206],{"type":40,"tag":136,"props":1198,"children":1199},{"class":138,"line":139},[1200],{"type":40,"tag":136,"props":1201,"children":1203},{"style":1202},"--shiki-light:#90A4AE;--shiki-light-font-style:italic;--shiki-default:#546E7A;--shiki-default-font-style:italic;--shiki-dark:#676E95;--shiki-dark-font-style:italic",[1204],{"type":45,"value":1205},"# Bad: Everything in one block\n",{"type":40,"tag":136,"props":1207,"children":1208},{"class":138,"line":148},[1209,1215,1221,1226,1232],{"type":40,"tag":136,"props":1210,"children":1212},{"style":1211},"--shiki-light:#E53935;--shiki-default:#F07178;--shiki-dark:#F07178",[1213],{"type":45,"value":1214},"agent_memory",{"type":40,"tag":136,"props":1216,"children":1218},{"style":1217},"--shiki-light:#39ADB5;--shiki-default:#89DDFF;--shiki-dark:#89DDFF",[1219],{"type":45,"value":1220},":",{"type":40,"tag":136,"props":1222,"children":1223},{"style":1217},[1224],{"type":45,"value":1225}," \"",{"type":40,"tag":136,"props":1227,"children":1229},{"style":1228},"--shiki-light:#91B859;--shiki-default:#C3E88D;--shiki-dark:#C3E88D",[1230],{"type":45,"value":1231},"Agent is helpful. User is John...",{"type":40,"tag":136,"props":1233,"children":1234},{"style":1217},[1235],{"type":45,"value":1236},"\"\n",{"type":40,"tag":48,"props":1238,"children":1239},{},[1240],{"type":45,"value":1241},"Split into focused blocks instead.",{"type":40,"tag":48,"props":1243,"children":1244},{},[1245,1250],{"type":40,"tag":317,"props":1246,"children":1247},{},[1248],{"type":45,"value":1249},"Too many memory blocks:",{"type":45,"value":1251},"\nCreating 10+ blocks when 3-4 would suffice. Start minimal, expand as needed.",{"type":40,"tag":48,"props":1253,"children":1254},{},[1255],{"type":40,"tag":317,"props":1256,"children":1257},{},[1258],{"type":45,"value":1259},"Poor descriptions:",{"type":40,"tag":124,"props":1261,"children":1263},{"className":1190,"code":1262,"language":1192,"meta":129,"style":129},"# Bad\ndata: \"Contains data\"\n",[1264],{"type":40,"tag":132,"props":1265,"children":1266},{"__ignoreMap":129},[1267,1275],{"type":40,"tag":136,"props":1268,"children":1269},{"class":138,"line":139},[1270],{"type":40,"tag":136,"props":1271,"children":1272},{"style":1202},[1273],{"type":45,"value":1274},"# Bad\n",{"type":40,"tag":136,"props":1276,"children":1277},{"class":138,"line":148},[1278,1283,1287,1291,1296],{"type":40,"tag":136,"props":1279,"children":1280},{"style":1211},[1281],{"type":45,"value":1282},"data",{"type":40,"tag":136,"props":1284,"children":1285},{"style":1217},[1286],{"type":45,"value":1220},{"type":40,"tag":136,"props":1288,"children":1289},{"style":1217},[1290],{"type":45,"value":1225},{"type":40,"tag":136,"props":1292,"children":1293},{"style":1228},[1294],{"type":45,"value":1295},"Contains data",{"type":40,"tag":136,"props":1297,"children":1298},{"style":1217},[1299],{"type":45,"value":1236},{"type":40,"tag":48,"props":1301,"children":1302},{},[1303,1305,1310],{"type":45,"value":1304},"Provide actionable guidance instead. See ",{"type":40,"tag":132,"props":1306,"children":1308},{"className":1307},[],[1309],{"type":45,"value":665},{"type":45,"value":373},{"type":40,"tag":48,"props":1312,"children":1313},{},[1314,1319],{"type":40,"tag":317,"props":1315,"children":1316},{},[1317],{"type":45,"value":1318},"Ignoring size limits:",{"type":45,"value":1320},"\nLetting blocks grow indefinitely until they hit limits. Monitor and manage proactively.",{"type":40,"tag":54,"props":1322,"children":1324},{"id":1323},"implementation-steps",[1325],{"type":45,"value":1326},"Implementation Steps",{"type":40,"tag":117,"props":1328,"children":1330},{"id":1329},"_1-design-phase",[1331],{"type":45,"value":1332},"1. Design Phase",{"type":40,"tag":66,"props":1334,"children":1335},{},[1336,1341,1346,1351],{"type":40,"tag":70,"props":1337,"children":1338},{},[1339],{"type":45,"value":1340},"Choose architecture based on requirements",{"type":40,"tag":70,"props":1342,"children":1343},{},[1344],{"type":45,"value":1345},"Design memory block structure",{"type":40,"tag":70,"props":1347,"children":1348},{},[1349],{"type":45,"value":1350},"Select appropriate model",{"type":40,"tag":70,"props":1352,"children":1353},{},[1354],{"type":45,"value":1355},"Plan tool configuration",{"type":40,"tag":117,"props":1357,"children":1359},{"id":1358},"_2-creation-phase-sdk",[1360],{"type":45,"value":1361},"2. Creation Phase (SDK)",{"type":40,"tag":48,"props":1363,"children":1364},{},[1365],{"type":40,"tag":317,"props":1366,"children":1367},{},[1368],{"type":45,"value":1369},"Python:",{"type":40,"tag":124,"props":1371,"children":1373},{"className":126,"code":1372,"language":128,"meta":129,"style":129},"from letta_client import Letta\n\nclient = Letta()  # Uses LETTA_API_KEY env var\n\n# Create agent with custom memory blocks\nagent = client.agents.create(\n    name=\"my-agent\",\n    model=\"openai\u002Fgpt-4o\",  # or \"anthropic\u002Fclaude-sonnet-4-20250514\"\n    embedding=\"openai\u002Ftext-embedding-3-small\",\n    memory_blocks=[\n        {\"label\": \"persona\", \"value\": \"You are a helpful assistant...\"},\n        {\"label\": \"human\", \"value\": \"User preferences and context...\"},\n        {\"label\": \"project\", \"value\": \"Current project details...\"},\n    ],\n    description=\"Agent for helping with X\",\n)\nprint(f\"Created agent: {agent.id}\")\n",[1374],{"type":40,"tag":132,"props":1375,"children":1376},{"__ignoreMap":129},[1377,1384,1391,1399,1406,1414,1421,1429,1437,1444,1451,1459,1467,1475,1482,1490,1497],{"type":40,"tag":136,"props":1378,"children":1379},{"class":138,"line":139},[1380],{"type":40,"tag":136,"props":1381,"children":1382},{},[1383],{"type":45,"value":145},{"type":40,"tag":136,"props":1385,"children":1386},{"class":138,"line":148},[1387],{"type":40,"tag":136,"props":1388,"children":1389},{"emptyLinePlaceholder":152},[1390],{"type":45,"value":155},{"type":40,"tag":136,"props":1392,"children":1393},{"class":138,"line":158},[1394],{"type":40,"tag":136,"props":1395,"children":1396},{},[1397],{"type":45,"value":1398},"client = Letta()  # Uses LETTA_API_KEY env var\n",{"type":40,"tag":136,"props":1400,"children":1401},{"class":138,"line":167},[1402],{"type":40,"tag":136,"props":1403,"children":1404},{"emptyLinePlaceholder":152},[1405],{"type":45,"value":155},{"type":40,"tag":136,"props":1407,"children":1408},{"class":138,"line":176},[1409],{"type":40,"tag":136,"props":1410,"children":1411},{},[1412],{"type":45,"value":1413},"# Create agent with custom memory blocks\n",{"type":40,"tag":136,"props":1415,"children":1416},{"class":138,"line":185},[1417],{"type":40,"tag":136,"props":1418,"children":1419},{},[1420],{"type":45,"value":173},{"type":40,"tag":136,"props":1422,"children":1423},{"class":138,"line":194},[1424],{"type":40,"tag":136,"props":1425,"children":1426},{},[1427],{"type":45,"value":1428},"    name=\"my-agent\",\n",{"type":40,"tag":136,"props":1430,"children":1431},{"class":138,"line":203},[1432],{"type":40,"tag":136,"props":1433,"children":1434},{},[1435],{"type":45,"value":1436},"    model=\"openai\u002Fgpt-4o\",  # or \"anthropic\u002Fclaude-sonnet-4-20250514\"\n",{"type":40,"tag":136,"props":1438,"children":1439},{"class":138,"line":212},[1440],{"type":40,"tag":136,"props":1441,"children":1442},{},[1443],{"type":45,"value":200},{"type":40,"tag":136,"props":1445,"children":1446},{"class":138,"line":221},[1447],{"type":40,"tag":136,"props":1448,"children":1449},{},[1450],{"type":45,"value":209},{"type":40,"tag":136,"props":1452,"children":1453},{"class":138,"line":230},[1454],{"type":40,"tag":136,"props":1455,"children":1456},{},[1457],{"type":45,"value":1458},"        {\"label\": \"persona\", \"value\": \"You are a helpful assistant...\"},\n",{"type":40,"tag":136,"props":1460,"children":1461},{"class":138,"line":239},[1462],{"type":40,"tag":136,"props":1463,"children":1464},{},[1465],{"type":45,"value":1466},"        {\"label\": \"human\", \"value\": \"User preferences and context...\"},\n",{"type":40,"tag":136,"props":1468,"children":1469},{"class":138,"line":248},[1470],{"type":40,"tag":136,"props":1471,"children":1472},{},[1473],{"type":45,"value":1474},"        {\"label\": \"project\", \"value\": \"Current project details...\"},\n",{"type":40,"tag":136,"props":1476,"children":1477},{"class":138,"line":256},[1478],{"type":40,"tag":136,"props":1479,"children":1480},{},[1481],{"type":45,"value":236},{"type":40,"tag":136,"props":1483,"children":1484},{"class":138,"line":265},[1485],{"type":40,"tag":136,"props":1486,"children":1487},{},[1488],{"type":45,"value":1489},"    description=\"Agent for helping with X\",\n",{"type":40,"tag":136,"props":1491,"children":1492},{"class":138,"line":274},[1493],{"type":40,"tag":136,"props":1494,"children":1495},{},[1496],{"type":45,"value":245},{"type":40,"tag":136,"props":1498,"children":1499},{"class":138,"line":283},[1500],{"type":40,"tag":136,"props":1501,"children":1502},{},[1503],{"type":45,"value":1504},"print(f\"Created agent: {agent.id}\")\n",{"type":40,"tag":48,"props":1506,"children":1507},{},[1508],{"type":40,"tag":317,"props":1509,"children":1510},{},[1511],{"type":45,"value":1512},"TypeScript:",{"type":40,"tag":124,"props":1514,"children":1518},{"className":1515,"code":1516,"language":1517,"meta":129,"style":129},"language-typescript shiki shiki-themes material-theme-lighter material-theme material-theme-palenight","import Letta from \"letta-client\";\n\nconst client = new Letta();\n\nconst agent = await client.agents.create({\n  name: \"my-agent\",\n  model: \"openai\u002Fgpt-4o\",\n  embedding: \"openai\u002Ftext-embedding-3-small\",\n  memoryBlocks: [\n    { label: \"persona\", value: \"You are a helpful assistant...\" },\n    { label: \"human\", value: \"User preferences and context...\" },\n    { label: \"project\", value: \"Current project details...\" },\n  ],\n  description: \"Agent for helping with X\",\n});\nconsole.log(`Created agent: ${agent.id}`);\n","typescript",[1519],{"type":40,"tag":132,"props":1520,"children":1521},{"__ignoreMap":129},[1522,1561,1568,1607,1614,1667,1697,1726,1755,1772,1833,1889,1946,1958,1987,2004],{"type":40,"tag":136,"props":1523,"children":1524},{"class":138,"line":139},[1525,1531,1537,1542,1546,1551,1556],{"type":40,"tag":136,"props":1526,"children":1528},{"style":1527},"--shiki-light:#39ADB5;--shiki-light-font-style:italic;--shiki-default:#89DDFF;--shiki-default-font-style:italic;--shiki-dark:#89DDFF;--shiki-dark-font-style:italic",[1529],{"type":45,"value":1530},"import",{"type":40,"tag":136,"props":1532,"children":1534},{"style":1533},"--shiki-light:#90A4AE;--shiki-default:#EEFFFF;--shiki-dark:#BABED8",[1535],{"type":45,"value":1536}," Letta ",{"type":40,"tag":136,"props":1538,"children":1539},{"style":1527},[1540],{"type":45,"value":1541},"from",{"type":40,"tag":136,"props":1543,"children":1544},{"style":1217},[1545],{"type":45,"value":1225},{"type":40,"tag":136,"props":1547,"children":1548},{"style":1228},[1549],{"type":45,"value":1550},"letta-client",{"type":40,"tag":136,"props":1552,"children":1553},{"style":1217},[1554],{"type":45,"value":1555},"\"",{"type":40,"tag":136,"props":1557,"children":1558},{"style":1217},[1559],{"type":45,"value":1560},";\n",{"type":40,"tag":136,"props":1562,"children":1563},{"class":138,"line":148},[1564],{"type":40,"tag":136,"props":1565,"children":1566},{"emptyLinePlaceholder":152},[1567],{"type":45,"value":155},{"type":40,"tag":136,"props":1569,"children":1570},{"class":138,"line":158},[1571,1577,1582,1587,1592,1598,1603],{"type":40,"tag":136,"props":1572,"children":1574},{"style":1573},"--shiki-light:#9C3EDA;--shiki-default:#C792EA;--shiki-dark:#C792EA",[1575],{"type":45,"value":1576},"const",{"type":40,"tag":136,"props":1578,"children":1579},{"style":1533},[1580],{"type":45,"value":1581}," client ",{"type":40,"tag":136,"props":1583,"children":1584},{"style":1217},[1585],{"type":45,"value":1586},"=",{"type":40,"tag":136,"props":1588,"children":1589},{"style":1217},[1590],{"type":45,"value":1591}," new",{"type":40,"tag":136,"props":1593,"children":1595},{"style":1594},"--shiki-light:#6182B8;--shiki-default:#82AAFF;--shiki-dark:#82AAFF",[1596],{"type":45,"value":1597}," Letta",{"type":40,"tag":136,"props":1599,"children":1600},{"style":1533},[1601],{"type":45,"value":1602},"()",{"type":40,"tag":136,"props":1604,"children":1605},{"style":1217},[1606],{"type":45,"value":1560},{"type":40,"tag":136,"props":1608,"children":1609},{"class":138,"line":167},[1610],{"type":40,"tag":136,"props":1611,"children":1612},{"emptyLinePlaceholder":152},[1613],{"type":45,"value":155},{"type":40,"tag":136,"props":1615,"children":1616},{"class":138,"line":176},[1617,1621,1626,1630,1635,1640,1644,1648,1652,1657,1662],{"type":40,"tag":136,"props":1618,"children":1619},{"style":1573},[1620],{"type":45,"value":1576},{"type":40,"tag":136,"props":1622,"children":1623},{"style":1533},[1624],{"type":45,"value":1625}," agent ",{"type":40,"tag":136,"props":1627,"children":1628},{"style":1217},[1629],{"type":45,"value":1586},{"type":40,"tag":136,"props":1631,"children":1632},{"style":1527},[1633],{"type":45,"value":1634}," await",{"type":40,"tag":136,"props":1636,"children":1637},{"style":1533},[1638],{"type":45,"value":1639}," client",{"type":40,"tag":136,"props":1641,"children":1642},{"style":1217},[1643],{"type":45,"value":373},{"type":40,"tag":136,"props":1645,"children":1646},{"style":1533},[1647],{"type":45,"value":22},{"type":40,"tag":136,"props":1649,"children":1650},{"style":1217},[1651],{"type":45,"value":373},{"type":40,"tag":136,"props":1653,"children":1654},{"style":1594},[1655],{"type":45,"value":1656},"create",{"type":40,"tag":136,"props":1658,"children":1659},{"style":1533},[1660],{"type":45,"value":1661},"(",{"type":40,"tag":136,"props":1663,"children":1664},{"style":1217},[1665],{"type":45,"value":1666},"{\n",{"type":40,"tag":136,"props":1668,"children":1669},{"class":138,"line":185},[1670,1675,1679,1683,1688,1692],{"type":40,"tag":136,"props":1671,"children":1672},{"style":1211},[1673],{"type":45,"value":1674},"  name",{"type":40,"tag":136,"props":1676,"children":1677},{"style":1217},[1678],{"type":45,"value":1220},{"type":40,"tag":136,"props":1680,"children":1681},{"style":1217},[1682],{"type":45,"value":1225},{"type":40,"tag":136,"props":1684,"children":1685},{"style":1228},[1686],{"type":45,"value":1687},"my-agent",{"type":40,"tag":136,"props":1689,"children":1690},{"style":1217},[1691],{"type":45,"value":1555},{"type":40,"tag":136,"props":1693,"children":1694},{"style":1217},[1695],{"type":45,"value":1696},",\n",{"type":40,"tag":136,"props":1698,"children":1699},{"class":138,"line":194},[1700,1705,1709,1713,1718,1722],{"type":40,"tag":136,"props":1701,"children":1702},{"style":1211},[1703],{"type":45,"value":1704},"  model",{"type":40,"tag":136,"props":1706,"children":1707},{"style":1217},[1708],{"type":45,"value":1220},{"type":40,"tag":136,"props":1710,"children":1711},{"style":1217},[1712],{"type":45,"value":1225},{"type":40,"tag":136,"props":1714,"children":1715},{"style":1228},[1716],{"type":45,"value":1717},"openai\u002Fgpt-4o",{"type":40,"tag":136,"props":1719,"children":1720},{"style":1217},[1721],{"type":45,"value":1555},{"type":40,"tag":136,"props":1723,"children":1724},{"style":1217},[1725],{"type":45,"value":1696},{"type":40,"tag":136,"props":1727,"children":1728},{"class":138,"line":203},[1729,1734,1738,1742,1747,1751],{"type":40,"tag":136,"props":1730,"children":1731},{"style":1211},[1732],{"type":45,"value":1733},"  embedding",{"type":40,"tag":136,"props":1735,"children":1736},{"style":1217},[1737],{"type":45,"value":1220},{"type":40,"tag":136,"props":1739,"children":1740},{"style":1217},[1741],{"type":45,"value":1225},{"type":40,"tag":136,"props":1743,"children":1744},{"style":1228},[1745],{"type":45,"value":1746},"openai\u002Ftext-embedding-3-small",{"type":40,"tag":136,"props":1748,"children":1749},{"style":1217},[1750],{"type":45,"value":1555},{"type":40,"tag":136,"props":1752,"children":1753},{"style":1217},[1754],{"type":45,"value":1696},{"type":40,"tag":136,"props":1756,"children":1757},{"class":138,"line":212},[1758,1763,1767],{"type":40,"tag":136,"props":1759,"children":1760},{"style":1211},[1761],{"type":45,"value":1762},"  memoryBlocks",{"type":40,"tag":136,"props":1764,"children":1765},{"style":1217},[1766],{"type":45,"value":1220},{"type":40,"tag":136,"props":1768,"children":1769},{"style":1533},[1770],{"type":45,"value":1771}," [\n",{"type":40,"tag":136,"props":1773,"children":1774},{"class":138,"line":221},[1775,1780,1785,1789,1793,1797,1801,1806,1811,1815,1819,1824,1828],{"type":40,"tag":136,"props":1776,"children":1777},{"style":1217},[1778],{"type":45,"value":1779},"    {",{"type":40,"tag":136,"props":1781,"children":1782},{"style":1211},[1783],{"type":45,"value":1784}," label",{"type":40,"tag":136,"props":1786,"children":1787},{"style":1217},[1788],{"type":45,"value":1220},{"type":40,"tag":136,"props":1790,"children":1791},{"style":1217},[1792],{"type":45,"value":1225},{"type":40,"tag":136,"props":1794,"children":1795},{"style":1228},[1796],{"type":45,"value":516},{"type":40,"tag":136,"props":1798,"children":1799},{"style":1217},[1800],{"type":45,"value":1555},{"type":40,"tag":136,"props":1802,"children":1803},{"style":1217},[1804],{"type":45,"value":1805},",",{"type":40,"tag":136,"props":1807,"children":1808},{"style":1211},[1809],{"type":45,"value":1810}," value",{"type":40,"tag":136,"props":1812,"children":1813},{"style":1217},[1814],{"type":45,"value":1220},{"type":40,"tag":136,"props":1816,"children":1817},{"style":1217},[1818],{"type":45,"value":1225},{"type":40,"tag":136,"props":1820,"children":1821},{"style":1228},[1822],{"type":45,"value":1823},"You are a helpful assistant...",{"type":40,"tag":136,"props":1825,"children":1826},{"style":1217},[1827],{"type":45,"value":1555},{"type":40,"tag":136,"props":1829,"children":1830},{"style":1217},[1831],{"type":45,"value":1832}," },\n",{"type":40,"tag":136,"props":1834,"children":1835},{"class":138,"line":230},[1836,1840,1844,1848,1852,1856,1860,1864,1868,1872,1876,1881,1885],{"type":40,"tag":136,"props":1837,"children":1838},{"style":1217},[1839],{"type":45,"value":1779},{"type":40,"tag":136,"props":1841,"children":1842},{"style":1211},[1843],{"type":45,"value":1784},{"type":40,"tag":136,"props":1845,"children":1846},{"style":1217},[1847],{"type":45,"value":1220},{"type":40,"tag":136,"props":1849,"children":1850},{"style":1217},[1851],{"type":45,"value":1225},{"type":40,"tag":136,"props":1853,"children":1854},{"style":1228},[1855],{"type":45,"value":527},{"type":40,"tag":136,"props":1857,"children":1858},{"style":1217},[1859],{"type":45,"value":1555},{"type":40,"tag":136,"props":1861,"children":1862},{"style":1217},[1863],{"type":45,"value":1805},{"type":40,"tag":136,"props":1865,"children":1866},{"style":1211},[1867],{"type":45,"value":1810},{"type":40,"tag":136,"props":1869,"children":1870},{"style":1217},[1871],{"type":45,"value":1220},{"type":40,"tag":136,"props":1873,"children":1874},{"style":1217},[1875],{"type":45,"value":1225},{"type":40,"tag":136,"props":1877,"children":1878},{"style":1228},[1879],{"type":45,"value":1880},"User preferences and context...",{"type":40,"tag":136,"props":1882,"children":1883},{"style":1217},[1884],{"type":45,"value":1555},{"type":40,"tag":136,"props":1886,"children":1887},{"style":1217},[1888],{"type":45,"value":1832},{"type":40,"tag":136,"props":1890,"children":1891},{"class":138,"line":239},[1892,1896,1900,1904,1908,1913,1917,1921,1925,1929,1933,1938,1942],{"type":40,"tag":136,"props":1893,"children":1894},{"style":1217},[1895],{"type":45,"value":1779},{"type":40,"tag":136,"props":1897,"children":1898},{"style":1211},[1899],{"type":45,"value":1784},{"type":40,"tag":136,"props":1901,"children":1902},{"style":1217},[1903],{"type":45,"value":1220},{"type":40,"tag":136,"props":1905,"children":1906},{"style":1217},[1907],{"type":45,"value":1225},{"type":40,"tag":136,"props":1909,"children":1910},{"style":1228},[1911],{"type":45,"value":1912},"project",{"type":40,"tag":136,"props":1914,"children":1915},{"style":1217},[1916],{"type":45,"value":1555},{"type":40,"tag":136,"props":1918,"children":1919},{"style":1217},[1920],{"type":45,"value":1805},{"type":40,"tag":136,"props":1922,"children":1923},{"style":1211},[1924],{"type":45,"value":1810},{"type":40,"tag":136,"props":1926,"children":1927},{"style":1217},[1928],{"type":45,"value":1220},{"type":40,"tag":136,"props":1930,"children":1931},{"style":1217},[1932],{"type":45,"value":1225},{"type":40,"tag":136,"props":1934,"children":1935},{"style":1228},[1936],{"type":45,"value":1937},"Current project details...",{"type":40,"tag":136,"props":1939,"children":1940},{"style":1217},[1941],{"type":45,"value":1555},{"type":40,"tag":136,"props":1943,"children":1944},{"style":1217},[1945],{"type":45,"value":1832},{"type":40,"tag":136,"props":1947,"children":1948},{"class":138,"line":248},[1949,1954],{"type":40,"tag":136,"props":1950,"children":1951},{"style":1533},[1952],{"type":45,"value":1953},"  ]",{"type":40,"tag":136,"props":1955,"children":1956},{"style":1217},[1957],{"type":45,"value":1696},{"type":40,"tag":136,"props":1959,"children":1960},{"class":138,"line":256},[1961,1966,1970,1974,1979,1983],{"type":40,"tag":136,"props":1962,"children":1963},{"style":1211},[1964],{"type":45,"value":1965},"  description",{"type":40,"tag":136,"props":1967,"children":1968},{"style":1217},[1969],{"type":45,"value":1220},{"type":40,"tag":136,"props":1971,"children":1972},{"style":1217},[1973],{"type":45,"value":1225},{"type":40,"tag":136,"props":1975,"children":1976},{"style":1228},[1977],{"type":45,"value":1978},"Agent for helping with X",{"type":40,"tag":136,"props":1980,"children":1981},{"style":1217},[1982],{"type":45,"value":1555},{"type":40,"tag":136,"props":1984,"children":1985},{"style":1217},[1986],{"type":45,"value":1696},{"type":40,"tag":136,"props":1988,"children":1989},{"class":138,"line":265},[1990,1995,2000],{"type":40,"tag":136,"props":1991,"children":1992},{"style":1217},[1993],{"type":45,"value":1994},"}",{"type":40,"tag":136,"props":1996,"children":1997},{"style":1533},[1998],{"type":45,"value":1999},")",{"type":40,"tag":136,"props":2001,"children":2002},{"style":1217},[2003],{"type":45,"value":1560},{"type":40,"tag":136,"props":2005,"children":2006},{"class":138,"line":274},[2007,2012,2016,2021,2025,2030,2035,2040,2045,2049,2054,2059,2063],{"type":40,"tag":136,"props":2008,"children":2009},{"style":1533},[2010],{"type":45,"value":2011},"console",{"type":40,"tag":136,"props":2013,"children":2014},{"style":1217},[2015],{"type":45,"value":373},{"type":40,"tag":136,"props":2017,"children":2018},{"style":1594},[2019],{"type":45,"value":2020},"log",{"type":40,"tag":136,"props":2022,"children":2023},{"style":1533},[2024],{"type":45,"value":1661},{"type":40,"tag":136,"props":2026,"children":2027},{"style":1217},[2028],{"type":45,"value":2029},"`",{"type":40,"tag":136,"props":2031,"children":2032},{"style":1228},[2033],{"type":45,"value":2034},"Created agent: ",{"type":40,"tag":136,"props":2036,"children":2037},{"style":1217},[2038],{"type":45,"value":2039},"${",{"type":40,"tag":136,"props":2041,"children":2042},{"style":1533},[2043],{"type":45,"value":2044},"agent",{"type":40,"tag":136,"props":2046,"children":2047},{"style":1217},[2048],{"type":45,"value":373},{"type":40,"tag":136,"props":2050,"children":2051},{"style":1533},[2052],{"type":45,"value":2053},"id",{"type":40,"tag":136,"props":2055,"children":2056},{"style":1217},[2057],{"type":45,"value":2058},"}`",{"type":40,"tag":136,"props":2060,"children":2061},{"style":1533},[2062],{"type":45,"value":1999},{"type":40,"tag":136,"props":2064,"children":2065},{"style":1217},[2066],{"type":45,"value":1560},{"type":40,"tag":48,"props":2068,"children":2069},{},[2070,2075,2077,2082,2084,2090,2092,2098,2100,2106],{"type":40,"tag":317,"props":2071,"children":2072},{},[2073],{"type":45,"value":2074},"Note:",{"type":45,"value":2076}," Letta Code CLI (",{"type":40,"tag":132,"props":2078,"children":2080},{"className":2079},[],[2081],{"type":45,"value":8},{"type":45,"value":2083}," command) creates agents interactively. Use ",{"type":40,"tag":132,"props":2085,"children":2087},{"className":2086},[],[2088],{"type":45,"value":2089},"letta --new-agent",{"type":45,"value":2091}," to start fresh, then ",{"type":40,"tag":132,"props":2093,"children":2095},{"className":2094},[],[2096],{"type":45,"value":2097},"\u002Frename",{"type":45,"value":2099}," and ",{"type":40,"tag":132,"props":2101,"children":2103},{"className":2102},[],[2104],{"type":45,"value":2105},"\u002Fdescription",{"type":45,"value":2107}," to configure.",{"type":40,"tag":117,"props":2109,"children":2111},{"id":2110},"_3-testing-phase",[2112],{"type":45,"value":2113},"3. Testing Phase",{"type":40,"tag":66,"props":2115,"children":2116},{},[2117,2122,2127],{"type":40,"tag":70,"props":2118,"children":2119},{},[2120],{"type":45,"value":2121},"Test with representative queries",{"type":40,"tag":70,"props":2123,"children":2124},{},[2125],{"type":45,"value":2126},"Monitor memory tool usage patterns",{"type":40,"tag":70,"props":2128,"children":2129},{},[2130],{"type":45,"value":2131},"Verify tool calling behavior",{"type":40,"tag":117,"props":2133,"children":2135},{"id":2134},"_4-iteration-phase",[2136],{"type":45,"value":2137},"4. Iteration Phase",{"type":40,"tag":66,"props":2139,"children":2140},{},[2141,2146,2151],{"type":40,"tag":70,"props":2142,"children":2143},{},[2144],{"type":45,"value":2145},"Refine memory block structure based on actual usage",{"type":40,"tag":70,"props":2147,"children":2148},{},[2149],{"type":45,"value":2150},"Optimize system instructions",{"type":40,"tag":70,"props":2152,"children":2153},{},[2154],{"type":45,"value":2155},"Adjust tool configurations",{"type":40,"tag":54,"props":2157,"children":2159},{"id":2158},"references",[2160],{"type":45,"value":2161},"References",{"type":40,"tag":48,"props":2163,"children":2164},{},[2165],{"type":45,"value":2166},"For detailed information on specific topics, consult the reference materials:",{"type":40,"tag":66,"props":2168,"children":2169},{},[2170,2180,2190,2200,2210,2220,2230,2240],{"type":40,"tag":70,"props":2171,"children":2172},{},[2173,2178],{"type":40,"tag":132,"props":2174,"children":2176},{"className":2175},[],[2177],{"type":45,"value":371},{"type":45,"value":2179}," - Architecture comparison and selection",{"type":40,"tag":70,"props":2181,"children":2182},{},[2183,2188],{"type":40,"tag":132,"props":2184,"children":2186},{"className":2185},[],[2187],{"type":45,"value":478},{"type":45,"value":2189}," - Memory types and when to use them",{"type":40,"tag":70,"props":2191,"children":2192},{},[2193,2198],{"type":40,"tag":132,"props":2194,"children":2196},{"className":2195},[],[2197],{"type":45,"value":657},{"type":45,"value":2199}," - Domain-specific memory block examples",{"type":40,"tag":70,"props":2201,"children":2202},{},[2203,2208],{"type":40,"tag":132,"props":2204,"children":2206},{"className":2205},[],[2207],{"type":45,"value":665},{"type":45,"value":2209}," - Writing effective block descriptions",{"type":40,"tag":70,"props":2211,"children":2212},{},[2213,2218],{"type":40,"tag":132,"props":2214,"children":2216},{"className":2215},[],[2217],{"type":45,"value":930},{"type":45,"value":2219}," - Managing memory block size limits",{"type":40,"tag":70,"props":2221,"children":2222},{},[2223,2228],{"type":40,"tag":132,"props":2224,"children":2226},{"className":2225},[],[2227],{"type":45,"value":1039},{"type":45,"value":2229}," - Multi-agent memory sharing patterns",{"type":40,"tag":70,"props":2231,"children":2232},{},[2233,2238],{"type":40,"tag":132,"props":2234,"children":2236},{"className":2235},[],[2237],{"type":45,"value":740},{"type":45,"value":2239}," - Model selection guidance",{"type":40,"tag":70,"props":2241,"children":2242},{},[2243,2248],{"type":40,"tag":132,"props":2244,"children":2246},{"className":2245},[],[2247],{"type":45,"value":819},{"type":45,"value":2249}," - Common tool configurations",{"type":40,"tag":2251,"props":2252,"children":2253},"style",{},[2254],{"type":45,"value":2255},"html .light .shiki span {color: var(--shiki-light);background: var(--shiki-light-bg);font-style: var(--shiki-light-font-style);font-weight: var(--shiki-light-font-weight);text-decoration: var(--shiki-light-text-decoration);}html.light .shiki span {color: var(--shiki-light);background: var(--shiki-light-bg);font-style: var(--shiki-light-font-style);font-weight: var(--shiki-light-font-weight);text-decoration: var(--shiki-light-text-decoration);}html .default .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}html.dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}",{"items":2257,"total":2373},[2258,2274,2291,2310,2323,2344,2354],{"slug":2259,"name":2259,"fn":2260,"description":2261,"org":2262,"tags":2263,"stars":23,"repoUrl":24,"updatedAt":2273},"1password","manage secrets with 1Password CLI","Set up and use 1Password CLI (op). Use when installing the CLI, enabling desktop app integration, signing in (single or multi-account), or reading\u002Finjecting\u002Frunning secrets via op.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[2264,2267,2270],{"name":2265,"slug":2266,"type":16},"Authentication","authentication",{"name":2268,"slug":2269,"type":16},"CLI","cli",{"name":2271,"slug":2272,"type":16},"Security","security","2026-07-13T06:24:39.504387",{"slug":2275,"name":2275,"fn":2276,"description":2277,"org":2278,"tags":2279,"stars":23,"repoUrl":24,"updatedAt":2290},"agent-slack","automate Slack messaging and workflows","Slack automation CLI — read\u002Fsend\u002Fsearch messages, browse threads and channels, manage channels, download attachments, look up users, and run workflows.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[2280,2283,2284,2287],{"name":2281,"slug":2282,"type":16},"Automation","automation",{"name":2268,"slug":2269,"type":16},{"name":2285,"slug":2286,"type":16},"Messaging","messaging",{"name":2288,"slug":2289,"type":16},"Slack","slack","2026-07-13T06:23:51.908511",{"slug":2292,"name":2292,"fn":2293,"description":2294,"org":2295,"tags":2296,"stars":23,"repoUrl":24,"updatedAt":2309},"ai-news","fetch and summarize AI news","Fetch and summarize recent AI news from curated RSS feeds (Hugging Face, VentureBeat, The Verge, OpenAI, Anthropic, DeepMind, etc.) and YouTube channels (Yannic Kilcher, Two Minute Papers, AI Explained, etc.). Also fetches full transcripts for specific YouTube videos. Use when the user asks about recent AI news, what's happened in AI lately, summaries of AI research or product announcements, or wants a digest of what's going on in the AI space.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[2297,2300,2303,2306],{"name":2298,"slug":2299,"type":16},"Communications","communications",{"name":2301,"slug":2302,"type":16},"LLM","llm",{"name":2304,"slug":2305,"type":16},"Research","research",{"name":2307,"slug":2308,"type":16},"Summarization","summarization","2026-07-13T06:24:20.520223",{"slug":2311,"name":2311,"fn":2312,"description":2313,"org":2314,"tags":2315,"stars":23,"repoUrl":24,"updatedAt":2322},"creating-letta-code-channels","build and debug Letta Code channels","Builds and debugs Letta Code channels, including first-party channel adapters and dynamic user channel plugins under ~\u002F.letta\u002Fchannels. Use when adding Telegram, WhatsApp, Bluesky, Slack, Discord, or custom channel support; testing channel routing, pairing, MessageChannel, runtime dependencies, or channel plugin manifests.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[2316,2317,2320,2321],{"name":21,"slug":22,"type":16},{"name":2318,"slug":2319,"type":16},"API Development","api-development",{"name":2285,"slug":2286,"type":16},{"name":2288,"slug":2289,"type":16},"2026-07-13T06:25:55.843495",{"slug":2324,"name":2324,"fn":2325,"description":2326,"org":2327,"tags":2328,"stars":23,"repoUrl":24,"updatedAt":2343},"datadog","query Datadog observability data","Query Datadog observability data (logs, metrics, monitors, dashboards, hosts) via direct API. Use when investigating production issues, checking monitors, searching logs, or accessing Datadog data.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[2329,2331,2334,2337,2340],{"name":2330,"slug":2324,"type":16},"Datadog",{"name":2332,"slug":2333,"type":16},"Logs","logs",{"name":2335,"slug":2336,"type":16},"Metrics","metrics",{"name":2338,"slug":2339,"type":16},"Monitoring","monitoring",{"name":2341,"slug":2342,"type":16},"Observability","observability","2026-07-13T06:24:27.990605",{"slug":2345,"name":2345,"fn":2346,"description":2347,"org":2348,"tags":2349,"stars":23,"repoUrl":24,"updatedAt":2353},"discord","automate Discord server and channel tasks","Discord automation CLI — send\u002Fread\u002Fsearch messages, manage channels and servers, react, create threads, pin messages, and look up users.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[2350,2351,2352],{"name":2281,"slug":2282,"type":16},{"name":2268,"slug":2269,"type":16},{"name":2285,"slug":2286,"type":16},"2026-07-13T06:24:26.62387",{"slug":2355,"name":2355,"fn":2356,"description":2357,"org":2358,"tags":2359,"stars":23,"repoUrl":24,"updatedAt":2372},"doc","create and edit Word documents","Use when the task involves reading, creating, or editing `.docx` documents, especially when formatting or layout fidelity matters; prefer `python-docx` plus the bundled `scripts\u002Frender_docx.py` for visual checks.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[2360,2363,2366,2369],{"name":2361,"slug":2362,"type":16},"Documents","documents",{"name":2364,"slug":2365,"type":16},"DOCX","docx",{"name":2367,"slug":2368,"type":16},"Office","office",{"name":2370,"slug":2371,"type":16},"Word","word","2026-07-13T06:23:44.299568",45,{"items":2375,"total":2530},[2376,2390,2405,2417,2429,2443,2453,2464,2476,2492,2503,2515],{"slug":2377,"name":2377,"fn":2378,"description":2379,"org":2380,"tags":2381,"stars":2387,"repoUrl":2388,"updatedAt":2389},"acquiring-skills","discover and install agent skills","Discover and install skills from Hermes, ClawHub, GitHub, and other registries. Load this skill whenever a user asks for a capability you don't already have — image generation, social media, email, calendar, finance, DevOps, search, browser automation, etc.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[2382,2383,2384],{"name":21,"slug":22,"type":16},{"name":2281,"slug":2282,"type":16},{"name":2385,"slug":2386,"type":16},"GitHub","github",2831,"https:\u002F\u002Fgithub.com\u002Fletta-ai\u002Fletta-code","2026-07-13T06:22:58.45767",{"slug":2391,"name":2392,"fn":2393,"description":2394,"org":2395,"tags":2396,"stars":2387,"repoUrl":2388,"updatedAt":2404},"context-doctor","Context Doctor","repair system prompt and memory degradation","Identify and repair degradation in system prompt, external memory, and skills preventing you from following instructions or remembering information as well as you should.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[2397,2398,2401],{"name":21,"slug":22,"type":16},{"name":2399,"slug":2400,"type":16},"AI Context","ai-context",{"name":2402,"slug":2403,"type":16},"Debugging","debugging","2026-07-13T06:22:50.151002",{"slug":2406,"name":2406,"fn":2407,"description":2408,"org":2409,"tags":2410,"stars":2387,"repoUrl":2388,"updatedAt":2416},"converting-mcps-to-skills","connect MCP servers to create skills","Connect to MCP (Model Context Protocol) servers and create skills for repeated use. Load when a user wants to use an MCP server, connect to external tools via MCP, or when they mention MCP, model context protocol, or specific MCP servers.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[2411,2412,2413],{"name":21,"slug":22,"type":16},{"name":2281,"slug":2282,"type":16},{"name":2414,"slug":2415,"type":16},"MCP","mcp","2026-07-13T06:23:37.646079",{"slug":2418,"name":2418,"fn":2419,"description":2420,"org":2421,"tags":2422,"stars":2387,"repoUrl":2388,"updatedAt":2428},"creating-mods","create and edit Letta Code mods","Creates and edits trusted local Letta Code mods, including tools, slash commands, local-only model providers, lifecycle\u002Fturn events, scoped conversation helpers, panels, and capability-gated behavior. Use when asked to make a mod, add an agent-callable tool, add a slash command, add a local provider\u002Fmodel adapter, transform turns, react to app events, or add lightweight mod UI outside the dedicated \u002Fstatusline flow.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[2423,2424,2425],{"name":21,"slug":22,"type":16},{"name":2281,"slug":2282,"type":16},{"name":2426,"slug":2427,"type":16},"Coding","coding","2026-07-23T05:42:38.133565",{"slug":2430,"name":2430,"fn":2431,"description":2432,"org":2433,"tags":2434,"stars":2387,"repoUrl":2388,"updatedAt":2442},"creating-skills","create and update agent skills","Guide for creating effective skills. This skill should be used when users want to create a new skill (or update an existing skill) that extends Letta Code's capabilities with specialized knowledge, workflows, or tool integrations.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[2435,2436,2439],{"name":21,"slug":22,"type":16},{"name":2437,"slug":2438,"type":16},"Documentation","documentation",{"name":2440,"slug":2441,"type":16},"Plugin Development","plugin-development","2026-07-13T06:22:56.998659",{"slug":2444,"name":2444,"fn":2445,"description":2446,"org":2447,"tags":2448,"stars":2387,"repoUrl":2388,"updatedAt":2452},"customizing-commands","create and manage Letta slash commands","Creates, edits, and enables Letta Code mod-provided slash commands. Use when the user asks to add a custom \u002Fcommand, slash command, command shortcut, scoped conversation-backed command, or command-driven panel behavior.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[2449,2450,2451],{"name":21,"slug":22,"type":16},{"name":2281,"slug":2282,"type":16},{"name":2268,"slug":2269,"type":16},"2026-07-13T06:23:18.266798",{"slug":2454,"name":2454,"fn":2455,"description":2456,"org":2457,"tags":2458,"stars":2387,"repoUrl":2388,"updatedAt":2463},"customizing-statusline","customize Letta Code statusline mods","Creates, edits, and migrates Letta Code statusline mods. Use when handling the \u002Fstatusline command or continuing work started by \u002Fstatusline.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[2459,2460],{"name":2268,"slug":2269,"type":16},{"name":2461,"slug":2462,"type":16},"Engineering","engineering","2026-07-13T06:23:27.465985",{"slug":2465,"name":2465,"fn":2466,"description":2467,"org":2468,"tags":2469,"stars":2387,"repoUrl":2388,"updatedAt":2475},"dispatching-coding-agents","dispatch stateless coding agents","Dispatch stateless coding agents (Claude Code or Codex) via Bash. Use when you're stuck, need a second opinion, or need parallel research on a hard problem. They have no memory — you must provide all context.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[2470,2471,2472],{"name":21,"slug":22,"type":16},{"name":2426,"slug":2427,"type":16},{"name":2473,"slug":2474,"type":16},"Multi-Agent","multi-agent","2026-07-26T05:46:56.388845",{"slug":2477,"name":2477,"fn":2478,"description":2479,"org":2480,"tags":2481,"stars":2387,"repoUrl":2388,"updatedAt":2491},"editing-letta-code-desktop-preferences","edit Letta Code Desktop preferences","Edits Letta Code Desktop (LCD) preferences by safely reading and updating ~\u002F.letta\u002Fdesktop_preferences.json. Use only when the user asks to change current Desktop\u002FLCD settings such as theme, default working directory, remote access preference, or remote environment name via the preferences JSON.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[2482,2485,2488],{"name":2483,"slug":2484,"type":16},"Configuration","configuration",{"name":2486,"slug":2487,"type":16},"Desktop","desktop",{"name":2489,"slug":2490,"type":16},"Themes","themes","2026-07-13T06:23:41.407811",{"slug":2493,"name":2493,"fn":2494,"description":2495,"org":2496,"tags":2497,"stars":2387,"repoUrl":2388,"updatedAt":2502},"finding-agents","locate and manage agents on server","Find other agents on the same server. Use when the user asks about other agents, wants to migrate memory from another agent, or needs to find an agent by name or tags.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[2498,2499],{"name":21,"slug":22,"type":16},{"name":2500,"slug":2501,"type":16},"Management","management","2026-07-16T06:02:17.297841",{"slug":2504,"name":2504,"fn":2505,"description":2506,"org":2507,"tags":2508,"stars":2387,"repoUrl":2388,"updatedAt":2514},"generating-mod-envs","generate Letta mod learning environments","Generates and reviews mod learning env JSON files for Letta Code local mods. Use when asked to teach, learn, or optimize a mod behavior; create, draft, validate, improve, or explain envs for `\u002Fmods learn --env`; or design evaluation scenarios, memory fixtures, requiredResultMarkers, requiredTraceMarkers, negative controls, and candidate diversity hints.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[2509,2510,2513],{"name":21,"slug":22,"type":16},{"name":2511,"slug":2512,"type":16},"AI Infrastructure","ai-infrastructure",{"name":2483,"slug":2484,"type":16},"2026-07-13T06:23:08.838181",{"slug":2516,"name":2516,"fn":2517,"description":2518,"org":2519,"tags":2520,"stars":2387,"repoUrl":2388,"updatedAt":2529},"image-generation","generate images from text prompts","Generate images from text prompts (and optionally edit\u002Fremix input images). Use when the user asks to create, generate, draw, render, or edit an image, illustration, logo, icon, diagram, or photo.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[2521,2524,2527],{"name":2522,"slug":2523,"type":16},"Creative","creative",{"name":2525,"slug":2526,"type":16},"Graphics","graphics",{"name":2528,"slug":2516,"type":16},"Image Generation","2026-07-13T06:23:06.189403",69]