[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill-aws-labs-finetuning":3,"mdc-s8oh10-key":42,"related-repo-aws-labs-finetuning":1215,"related-org-aws-labs-finetuning":1322},{"slug":4,"name":4,"fn":5,"description":6,"org":7,"tags":12,"stars":26,"repoUrl":27,"updatedAt":28,"license":29,"forks":30,"topics":31,"repo":37,"sourceUrl":40,"mdContent":41},"finetuning","fine-tune models with SageMaker","Generates code that fine-tunes a base model using SageMaker serverless training jobs. Use when the user says \"start training\", \"fine-tune my model\", \"I'm ready to train\", or when the plan reaches the finetuning step. Supports SFT, DPO, RLVR, and RLAIF trainers, including RLVR Lambda reward function and RLAIF custom prompt creation.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},"aws-labs","AWS Labs","https:\u002F\u002Fpexgzepcugksgbtrxkhf.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Forg-logos\u002Faws-labs.png","awslabs",[13,17,20,23],{"name":14,"slug":15,"type":16},"LLM","llm","tag",{"name":18,"slug":19,"type":16},"Machine Learning","machine-learning",{"name":21,"slug":22,"type":16},"AI Infrastructure","ai-infrastructure",{"name":24,"slug":25,"type":16},"AWS","aws",831,"https:\u002F\u002Fgithub.com\u002Fawslabs\u002Fagent-plugins","2026-07-12T08:39:37.178051",null,127,[32,33,34,25,35,36],"agent-plugins","agent-skills","agents","coding-agent-skills","coding-agents",{"repoUrl":27,"stars":26,"forks":30,"topics":38,"description":39},[32,33,34,25,35,36],"Agent Plugins for AWS equip AI coding agents with the skills to help you architect, deploy, and operate on AWS.","https:\u002F\u002Fgithub.com\u002Fawslabs\u002Fagent-plugins\u002Ftree\u002FHEAD\u002Fplugins\u002Fsagemaker-ai\u002Fskills\u002Ffinetuning","---\nname: finetuning\ndescription: Generates code that fine-tunes a base model using SageMaker serverless training jobs. Use when the user says \"start training\", \"fine-tune my model\", \"I'm ready to train\", or when the plan reaches the finetuning step. Supports SFT, DPO, RLVR, and RLAIF trainers, including RLVR Lambda reward function and RLAIF custom prompt creation.\nmetadata:\n  version: \"1.0.0\"\n---\n\n# Prerequisites\n\nBefore starting this workflow, verify:\n\n1. A `use_case_spec.md` file exists\n   - If missing: Activate the `use-case-specification` skill first, then resume\n   - DON'T EVER offer to create a use case spec without activating the use-case-specification skill.\n\n2. A fine-tuning technique (SFT, DPO, RLVR, RLAIF, or CPT\u002FRFT (for Nova)) and base model have already been selected\n   - If missing: Activate the `model-selection` and\u002For `finetuning-technique` skills to collect what's missing, then resume\n   - Don't make recommendations on the spot. You MUST activate the appropriate skill.\n\n3. A base model name available on SageMakerHub has been identified\n   - If missing: Activate the `model-selection` skill to get it\n   - **Important:** Only use the model name that `model-selection` retrieves, as it may differ from other commonly used names for the same model\n\n4. The SDK environment has been verified (SDK version, region, execution role)\n   - If not done: Activate the `sdk-getting-started` skill first, then resume\n\n5. A training dataset uploaded to a bucket in the environment's default region.\n   - If not met: Help the user upload the dataset to the correct S3\n\n---\n\n# Critical Rules\n\n## Code Generation Rules\n\n- ✅ Use EXACTLY the imports shown in each code template\n- ❌ Do NOT add additional imports even if they seem helpful\n- ❌ Do NOT create variables before they're needed in that section\n- 📋 Copy the code structure precisely - no improvisation\n- 🎯 Follow the minimal code principle strictly\n- ✅ When writing code, make sure the indentation and f strings are correct\n\n## User Communication Rules\n\n- ❌ NEVER offer to move on to a downstream skill while training is in progress (logically impossible)\n- ❌ NEVER set ACCEPT_EULA to True without explicit user confirmation in the conversation\n- ✅ Always mention both the number AND title of sections you reference\n- ✅ If user asks how to run (notebook): If `run_cell` is available, offer to run it. Otherwise, tell them to run cells one by one (mention ipykernel requirement).\n- ✅ If user asks how to run (script): Tell them to run with `python3 \u003Cscript>.py`\n\n---\n\n# Workflow\n\n## 1. Code Generation Setup\n\n### 1.1 Directory Setup\n\n1. Identify project directory from conversation context\n   - If unclear (multiple relevant directories exist) → Ask user which folder to use\n   - If no project directory exists → activate the **directory-management** skill to set one up\n\n⏸ Wait for user.\n\n### 1.2 Select Code Template\n\nRead `references\u002Fcode_output_guide.md` for output format rules, then read the code template matching the finetuning strategy:\n\n- SFT → `code_templates\u002Fsft.py`\n- DPO → `code_templates\u002Fdpo.py`\n- RLVR → `code_templates\u002Frlvr.py`\n- RLAIF with built-in rewards → `code_templates\u002Frlaif_builtin.py`\n- RLAIF with custom prompt → `code_templates\u002Frlaif_custom_prompt.py`\n\nThe template is a Python file where each `# Cell N: Label` comment marks the start of a new section. Split on these markers — everything between one marker and the next becomes one unit of output.\n\n### 1.3 Generate Code\n\n1. Write the code from the template following the rules in `code_output_guide.md`\n2. Use same order, dependencies, and imports as the template\n3. DO NOT improvise or add extra code\n4. If the model is **NOT** a Meta\u002FLlama model (model ID does NOT start with `meta-`):\n   - Omit the `ACCEPT_EULA = False` line from the config cell\n   - Omit the `accept_eula=ACCEPT_EULA,` line from the trainer call\n5. If the model is from the Nova family, omit any code containing `max_epochs` or `lr_warmup_steps_ratio` from the Configure Trainer section and the Hyperparameter Overrides section\n\n### 1.4 Auto-Generate Configuration Values\n\n**In the 'Setup & Credentials' cell, populate:**\n\n1. **BASE_MODEL**\n   - Use the exact SageMakerHub model name from context\n\n2. **MODEL_PACKAGE_GROUP_NAME**\n   - Generate from use case (read `use_case_spec.md` if needed)\n   - Format rules:\n     - Lowercase, alphanumeric with hyphens only\n     - 1-63 characters\n     - Pattern: `[a-zA-Z0-9](-*[a-zA-Z0-9]){0,62}`\n     - Example: \"Customer Support Chatbot\" → `customer-support-chatbot-v1`\n\n3. Save notebook\n\n## 2. RLVR Reward Function (for RLVR only, skip this section if technique is SFT or DPO)\n\n### 2.1 Check Reward Function Status\n\n- Ask if user has a reward function already, or would like help creating one.\n  - If user says they have one → Ask for the SageMaker Hub Evaluator ARN. Only proceed to Section 2.3 once the user provides a valid Evaluator ARN. If they don't have it registered as a SageMaker Hub Evaluator, continue to 2.2.\n  - If user says they do not have one → Continue to 2.2\n\n### 2.2 Generate Reward Function From Template\n\n1. Follow workflow in `references\u002Frlvr_reward_function.md` section \"Helping Users Create Custom Reward Functions\"\n\n### 2.3 Set CUSTOM_REWARD_FUNCTION value\n\n1. Set the value for `CUSTOM_REWARD_FUNCTION` in the Notebook with the ARN of the reward function (either given directly by the user, or from the function generation code as `evaluator.arn`).\n\n## 3. RLAIF (for RLAIF only, skip this section if technique is not RLAIF)\n\nRead `references\u002Frlaif_guide.md` and follow its instructions.\n\n## 4. EULA review and acceptance\n\n1. Look up the official license link for the selected base model from references\u002Feula_links.md\n2. Display the license to the user following the phrasing in references\u002Feula_links.md. For OSS models: \"This model is licensed under **{License}**. Please review the license terms here: {URL}.\" For Nova models: \"This model is subject to the AWS Service Terms: {URL}.\"\n3. Check if the selected base model is a Meta\u002FLlama model (model ID starts with `meta-`)\n   - **If Meta\u002FLlama**: Tell the user they must read and agree to the EULA before using this model. Ask: \"Do you accept the license terms? (yes\u002Fno)\". If the user confirms, set `ACCEPT_EULA = True` and uncomment `accept_eula=ACCEPT_EULA` in the generated notebook. If the user declines, leave `ACCEPT_EULA = False` and warn that training will fail without acceptance.\n   - **If non-Meta**: Inform the user of the license for their awareness. No code-level action needed — the `ACCEPT_EULA` variable and `accept_eula` parameter should already be omitted from the notebook (see Step 1.3).\n\n## 5. Post-Generation\n\nAfter generating the code, offer to run it. Training can take hours depending on your dataset and model.\n\n**Notebook mode:** If `run_cell` is available, offer to run the cells. Otherwise tell the user to run cells themselves.\n\n**Script mode:** Present the user with options:\n\n> \"Would you like me to:\n>\n> 1. Leave it to you — run with `python scripts\u002F[script_name]`\n> 2. Run it and wait until it's done\n> 3. Start it but don't wait — we can check status later\"\n\n- **Option 1:** Done. Wait for user to come back.\n- **Option 2:** Execute the script as-is. `trainer.train(wait=True)` blocks until complete. Report final status.\n- **Option 3:** Change `wait=True` to `wait=False` in the script, execute, report the training job name.\n\n**Checking status:**\n\n- `describe-training-job --training-job-name NAME` → `TrainingJobStatus`, `FailureReason`, `SecondaryStatusTransitions`\n- For model package ARN after completion: `list-model-packages --model-package-group-name GROUP_NAME --sort-by CreationTime --sort-order Descending --max-results 1`\n\n**Showing results after completion:**\n\n- Use `scripts\u002Fmlflow_reference.py` as the pattern to query MLflow metrics\n- Present loss by epoch as a text table (total_loss, val_eval_total_loss for SFT; rewards\u002Fmargins for DPO; critic\u002Frewards\u002Fmean for RLVR)\n\n**CRITICAL:**\n\n- DON'T suggest moving to next steps before training completes\n- DON'T elaborate on the next steps unless the user specifically asks you about them.\n\n## 6. Continuous Customization\n\nIf the user wants to finetune a model they had already customized, follow the instructions in references\u002Fcontinuous_customization.md\n\n---\n\n# References\n\n- `rlvr_reward_function.md` - Lambda reward function creation guide (RLVR only)\n- `templates\u002Frlvr_reward_function_source_template.py` - Lambda reward function source template for open-weights models (RLVR only)\n- `templates\u002Fnova_rlvr_reward_function_source_template.py` - Lambda reward function source template for Nova 2.0 Lite (RLVR only)\n- `code_templates\u002Fsft.py` - Complete notebook template for Supervised Fine-Tuning (OSS path)\n- `code_templates\u002Fdpo.py` - Complete notebook template for Direct Preference Optimization (OSS path)\n- `code_templates\u002Frlvr.py` - Complete notebook template for Reinforcement Learning from Verifiable Rewards (OSS path)\n- `references\u002Fcontinuous_customization.md` - Instructions on fine-tuning an already fine-tuned model.\n- `rlaif_guide.md` - instructions on RLAIF finetuning options\n- `rlaif_builtin.py` - Code template for RLAIF with built-in judge prompt\n- `rlaif_custom_prompt.py` - Code template for RLAIF with custom judge prompt\n",{"data":43,"body":46},{"name":4,"description":6,"metadata":44},{"version":45},"1.0.0",{"type":47,"children":48},"root",[49,58,64,208,212,218,225,258,264,306,309,315,321,328,356,361,367,380,438,451,457,550,556,564,651,657,663,684,690,706,712,736,742,754,760,853,859,864,881,891,924,981,989,1035,1043,1064,1072,1085,1091,1096,1099,1105],{"type":50,"tag":51,"props":52,"children":54},"element","h1",{"id":53},"prerequisites",[55],{"type":56,"value":57},"text","Prerequisites",{"type":50,"tag":59,"props":60,"children":61},"p",{},[62],{"type":56,"value":63},"Before starting this workflow, verify:",{"type":50,"tag":65,"props":66,"children":67},"ol",{},[68,105,138,175,195],{"type":50,"tag":69,"props":70,"children":71},"li",{},[72,74,81,83],{"type":56,"value":73},"A ",{"type":50,"tag":75,"props":76,"children":78},"code",{"className":77},[],[79],{"type":56,"value":80},"use_case_spec.md",{"type":56,"value":82}," file exists",{"type":50,"tag":84,"props":85,"children":86},"ul",{},[87,100],{"type":50,"tag":69,"props":88,"children":89},{},[90,92,98],{"type":56,"value":91},"If missing: Activate the ",{"type":50,"tag":75,"props":93,"children":95},{"className":94},[],[96],{"type":56,"value":97},"use-case-specification",{"type":56,"value":99}," skill first, then resume",{"type":50,"tag":69,"props":101,"children":102},{},[103],{"type":56,"value":104},"DON'T EVER offer to create a use case spec without activating the use-case-specification skill.",{"type":50,"tag":69,"props":106,"children":107},{},[108,110],{"type":56,"value":109},"A fine-tuning technique (SFT, DPO, RLVR, RLAIF, or CPT\u002FRFT (for Nova)) and base model have already been selected",{"type":50,"tag":84,"props":111,"children":112},{},[113,133],{"type":50,"tag":69,"props":114,"children":115},{},[116,117,123,125,131],{"type":56,"value":91},{"type":50,"tag":75,"props":118,"children":120},{"className":119},[],[121],{"type":56,"value":122},"model-selection",{"type":56,"value":124}," and\u002For ",{"type":50,"tag":75,"props":126,"children":128},{"className":127},[],[129],{"type":56,"value":130},"finetuning-technique",{"type":56,"value":132}," skills to collect what's missing, then resume",{"type":50,"tag":69,"props":134,"children":135},{},[136],{"type":56,"value":137},"Don't make recommendations on the spot. You MUST activate the appropriate skill.",{"type":50,"tag":69,"props":139,"children":140},{},[141,143],{"type":56,"value":142},"A base model name available on SageMakerHub has been identified",{"type":50,"tag":84,"props":144,"children":145},{},[146,157],{"type":50,"tag":69,"props":147,"children":148},{},[149,150,155],{"type":56,"value":91},{"type":50,"tag":75,"props":151,"children":153},{"className":152},[],[154],{"type":56,"value":122},{"type":56,"value":156}," skill to get it",{"type":50,"tag":69,"props":158,"children":159},{},[160,166,168,173],{"type":50,"tag":161,"props":162,"children":163},"strong",{},[164],{"type":56,"value":165},"Important:",{"type":56,"value":167}," Only use the model name that ",{"type":50,"tag":75,"props":169,"children":171},{"className":170},[],[172],{"type":56,"value":122},{"type":56,"value":174}," retrieves, as it may differ from other commonly used names for the same model",{"type":50,"tag":69,"props":176,"children":177},{},[178,180],{"type":56,"value":179},"The SDK environment has been verified (SDK version, region, execution role)",{"type":50,"tag":84,"props":181,"children":182},{},[183],{"type":50,"tag":69,"props":184,"children":185},{},[186,188,194],{"type":56,"value":187},"If not done: Activate the ",{"type":50,"tag":75,"props":189,"children":191},{"className":190},[],[192],{"type":56,"value":193},"sdk-getting-started",{"type":56,"value":99},{"type":50,"tag":69,"props":196,"children":197},{},[198,200],{"type":56,"value":199},"A training dataset uploaded to a bucket in the environment's default region.",{"type":50,"tag":84,"props":201,"children":202},{},[203],{"type":50,"tag":69,"props":204,"children":205},{},[206],{"type":56,"value":207},"If not met: Help the user upload the dataset to the correct S3",{"type":50,"tag":209,"props":210,"children":211},"hr",{},[],{"type":50,"tag":51,"props":213,"children":215},{"id":214},"critical-rules",[216],{"type":56,"value":217},"Critical Rules",{"type":50,"tag":219,"props":220,"children":222},"h2",{"id":221},"code-generation-rules",[223],{"type":56,"value":224},"Code Generation Rules",{"type":50,"tag":84,"props":226,"children":227},{},[228,233,238,243,248,253],{"type":50,"tag":69,"props":229,"children":230},{},[231],{"type":56,"value":232},"✅ Use EXACTLY the imports shown in each code template",{"type":50,"tag":69,"props":234,"children":235},{},[236],{"type":56,"value":237},"❌ Do NOT add additional imports even if they seem helpful",{"type":50,"tag":69,"props":239,"children":240},{},[241],{"type":56,"value":242},"❌ Do NOT create variables before they're needed in that section",{"type":50,"tag":69,"props":244,"children":245},{},[246],{"type":56,"value":247},"📋 Copy the code structure precisely - no improvisation",{"type":50,"tag":69,"props":249,"children":250},{},[251],{"type":56,"value":252},"🎯 Follow the minimal code principle strictly",{"type":50,"tag":69,"props":254,"children":255},{},[256],{"type":56,"value":257},"✅ When writing code, make sure the indentation and f strings are correct",{"type":50,"tag":219,"props":259,"children":261},{"id":260},"user-communication-rules",[262],{"type":56,"value":263},"User Communication Rules",{"type":50,"tag":84,"props":265,"children":266},{},[267,272,277,282,295],{"type":50,"tag":69,"props":268,"children":269},{},[270],{"type":56,"value":271},"❌ NEVER offer to move on to a downstream skill while training is in progress (logically impossible)",{"type":50,"tag":69,"props":273,"children":274},{},[275],{"type":56,"value":276},"❌ NEVER set ACCEPT_EULA to True without explicit user confirmation in the conversation",{"type":50,"tag":69,"props":278,"children":279},{},[280],{"type":56,"value":281},"✅ Always mention both the number AND title of sections you reference",{"type":50,"tag":69,"props":283,"children":284},{},[285,287,293],{"type":56,"value":286},"✅ If user asks how to run (notebook): If ",{"type":50,"tag":75,"props":288,"children":290},{"className":289},[],[291],{"type":56,"value":292},"run_cell",{"type":56,"value":294}," is available, offer to run it. Otherwise, tell them to run cells one by one (mention ipykernel requirement).",{"type":50,"tag":69,"props":296,"children":297},{},[298,300],{"type":56,"value":299},"✅ If user asks how to run (script): Tell them to run with ",{"type":50,"tag":75,"props":301,"children":303},{"className":302},[],[304],{"type":56,"value":305},"python3 \u003Cscript>.py",{"type":50,"tag":209,"props":307,"children":308},{},[],{"type":50,"tag":51,"props":310,"children":312},{"id":311},"workflow",[313],{"type":56,"value":314},"Workflow",{"type":50,"tag":219,"props":316,"children":318},{"id":317},"_1-code-generation-setup",[319],{"type":56,"value":320},"1. Code Generation Setup",{"type":50,"tag":322,"props":323,"children":325},"h3",{"id":324},"_11-directory-setup",[326],{"type":56,"value":327},"1.1 Directory Setup",{"type":50,"tag":65,"props":329,"children":330},{},[331],{"type":50,"tag":69,"props":332,"children":333},{},[334,336],{"type":56,"value":335},"Identify project directory from conversation context\n",{"type":50,"tag":84,"props":337,"children":338},{},[339,344],{"type":50,"tag":69,"props":340,"children":341},{},[342],{"type":56,"value":343},"If unclear (multiple relevant directories exist) → Ask user which folder to use",{"type":50,"tag":69,"props":345,"children":346},{},[347,349,354],{"type":56,"value":348},"If no project directory exists → activate the ",{"type":50,"tag":161,"props":350,"children":351},{},[352],{"type":56,"value":353},"directory-management",{"type":56,"value":355}," skill to set one up",{"type":50,"tag":59,"props":357,"children":358},{},[359],{"type":56,"value":360},"⏸ Wait for user.",{"type":50,"tag":322,"props":362,"children":364},{"id":363},"_12-select-code-template",[365],{"type":56,"value":366},"1.2 Select Code Template",{"type":50,"tag":59,"props":368,"children":369},{},[370,372,378],{"type":56,"value":371},"Read ",{"type":50,"tag":75,"props":373,"children":375},{"className":374},[],[376],{"type":56,"value":377},"references\u002Fcode_output_guide.md",{"type":56,"value":379}," for output format rules, then read the code template matching the finetuning strategy:",{"type":50,"tag":84,"props":381,"children":382},{},[383,394,405,416,427],{"type":50,"tag":69,"props":384,"children":385},{},[386,388],{"type":56,"value":387},"SFT → ",{"type":50,"tag":75,"props":389,"children":391},{"className":390},[],[392],{"type":56,"value":393},"code_templates\u002Fsft.py",{"type":50,"tag":69,"props":395,"children":396},{},[397,399],{"type":56,"value":398},"DPO → ",{"type":50,"tag":75,"props":400,"children":402},{"className":401},[],[403],{"type":56,"value":404},"code_templates\u002Fdpo.py",{"type":50,"tag":69,"props":406,"children":407},{},[408,410],{"type":56,"value":409},"RLVR → ",{"type":50,"tag":75,"props":411,"children":413},{"className":412},[],[414],{"type":56,"value":415},"code_templates\u002Frlvr.py",{"type":50,"tag":69,"props":417,"children":418},{},[419,421],{"type":56,"value":420},"RLAIF with built-in rewards → ",{"type":50,"tag":75,"props":422,"children":424},{"className":423},[],[425],{"type":56,"value":426},"code_templates\u002Frlaif_builtin.py",{"type":50,"tag":69,"props":428,"children":429},{},[430,432],{"type":56,"value":431},"RLAIF with custom prompt → ",{"type":50,"tag":75,"props":433,"children":435},{"className":434},[],[436],{"type":56,"value":437},"code_templates\u002Frlaif_custom_prompt.py",{"type":50,"tag":59,"props":439,"children":440},{},[441,443,449],{"type":56,"value":442},"The template is a Python file where each ",{"type":50,"tag":75,"props":444,"children":446},{"className":445},[],[447],{"type":56,"value":448},"# Cell N: Label",{"type":56,"value":450}," comment marks the start of a new section. Split on these markers — everything between one marker and the next becomes one unit of output.",{"type":50,"tag":322,"props":452,"children":454},{"id":453},"_13-generate-code",[455],{"type":56,"value":456},"1.3 Generate Code",{"type":50,"tag":65,"props":458,"children":459},{},[460,471,476,481,529],{"type":50,"tag":69,"props":461,"children":462},{},[463,465],{"type":56,"value":464},"Write the code from the template following the rules in ",{"type":50,"tag":75,"props":466,"children":468},{"className":467},[],[469],{"type":56,"value":470},"code_output_guide.md",{"type":50,"tag":69,"props":472,"children":473},{},[474],{"type":56,"value":475},"Use same order, dependencies, and imports as the template",{"type":50,"tag":69,"props":477,"children":478},{},[479],{"type":56,"value":480},"DO NOT improvise or add extra code",{"type":50,"tag":69,"props":482,"children":483},{},[484,486,491,493,499,501],{"type":56,"value":485},"If the model is ",{"type":50,"tag":161,"props":487,"children":488},{},[489],{"type":56,"value":490},"NOT",{"type":56,"value":492}," a Meta\u002FLlama model (model ID does NOT start with ",{"type":50,"tag":75,"props":494,"children":496},{"className":495},[],[497],{"type":56,"value":498},"meta-",{"type":56,"value":500},"):\n",{"type":50,"tag":84,"props":502,"children":503},{},[504,517],{"type":50,"tag":69,"props":505,"children":506},{},[507,509,515],{"type":56,"value":508},"Omit the ",{"type":50,"tag":75,"props":510,"children":512},{"className":511},[],[513],{"type":56,"value":514},"ACCEPT_EULA = False",{"type":56,"value":516}," line from the config cell",{"type":50,"tag":69,"props":518,"children":519},{},[520,521,527],{"type":56,"value":508},{"type":50,"tag":75,"props":522,"children":524},{"className":523},[],[525],{"type":56,"value":526},"accept_eula=ACCEPT_EULA,",{"type":56,"value":528}," line from the trainer call",{"type":50,"tag":69,"props":530,"children":531},{},[532,534,540,542,548],{"type":56,"value":533},"If the model is from the Nova family, omit any code containing ",{"type":50,"tag":75,"props":535,"children":537},{"className":536},[],[538],{"type":56,"value":539},"max_epochs",{"type":56,"value":541}," or ",{"type":50,"tag":75,"props":543,"children":545},{"className":544},[],[546],{"type":56,"value":547},"lr_warmup_steps_ratio",{"type":56,"value":549}," from the Configure Trainer section and the Hyperparameter Overrides section",{"type":50,"tag":322,"props":551,"children":553},{"id":552},"_14-auto-generate-configuration-values",[554],{"type":56,"value":555},"1.4 Auto-Generate Configuration Values",{"type":50,"tag":59,"props":557,"children":558},{},[559],{"type":50,"tag":161,"props":560,"children":561},{},[562],{"type":56,"value":563},"In the 'Setup & Credentials' cell, populate:",{"type":50,"tag":65,"props":565,"children":566},{},[567,583,646],{"type":50,"tag":69,"props":568,"children":569},{},[570,575],{"type":50,"tag":161,"props":571,"children":572},{},[573],{"type":56,"value":574},"BASE_MODEL",{"type":50,"tag":84,"props":576,"children":577},{},[578],{"type":50,"tag":69,"props":579,"children":580},{},[581],{"type":56,"value":582},"Use the exact SageMakerHub model name from context",{"type":50,"tag":69,"props":584,"children":585},{},[586,591],{"type":50,"tag":161,"props":587,"children":588},{},[589],{"type":56,"value":590},"MODEL_PACKAGE_GROUP_NAME",{"type":50,"tag":84,"props":592,"children":593},{},[594,606],{"type":50,"tag":69,"props":595,"children":596},{},[597,599,604],{"type":56,"value":598},"Generate from use case (read ",{"type":50,"tag":75,"props":600,"children":602},{"className":601},[],[603],{"type":56,"value":80},{"type":56,"value":605}," if needed)",{"type":50,"tag":69,"props":607,"children":608},{},[609,611],{"type":56,"value":610},"Format rules:\n",{"type":50,"tag":84,"props":612,"children":613},{},[614,619,624,635],{"type":50,"tag":69,"props":615,"children":616},{},[617],{"type":56,"value":618},"Lowercase, alphanumeric with hyphens only",{"type":50,"tag":69,"props":620,"children":621},{},[622],{"type":56,"value":623},"1-63 characters",{"type":50,"tag":69,"props":625,"children":626},{},[627,629],{"type":56,"value":628},"Pattern: ",{"type":50,"tag":75,"props":630,"children":632},{"className":631},[],[633],{"type":56,"value":634},"[a-zA-Z0-9](-*[a-zA-Z0-9]){0,62}",{"type":50,"tag":69,"props":636,"children":637},{},[638,640],{"type":56,"value":639},"Example: \"Customer Support Chatbot\" → ",{"type":50,"tag":75,"props":641,"children":643},{"className":642},[],[644],{"type":56,"value":645},"customer-support-chatbot-v1",{"type":50,"tag":69,"props":647,"children":648},{},[649],{"type":56,"value":650},"Save notebook",{"type":50,"tag":219,"props":652,"children":654},{"id":653},"_2-rlvr-reward-function-for-rlvr-only-skip-this-section-if-technique-is-sft-or-dpo",[655],{"type":56,"value":656},"2. RLVR Reward Function (for RLVR only, skip this section if technique is SFT or DPO)",{"type":50,"tag":322,"props":658,"children":660},{"id":659},"_21-check-reward-function-status",[661],{"type":56,"value":662},"2.1 Check Reward Function Status",{"type":50,"tag":84,"props":664,"children":665},{},[666],{"type":50,"tag":69,"props":667,"children":668},{},[669,671],{"type":56,"value":670},"Ask if user has a reward function already, or would like help creating one.\n",{"type":50,"tag":84,"props":672,"children":673},{},[674,679],{"type":50,"tag":69,"props":675,"children":676},{},[677],{"type":56,"value":678},"If user says they have one → Ask for the SageMaker Hub Evaluator ARN. Only proceed to Section 2.3 once the user provides a valid Evaluator ARN. If they don't have it registered as a SageMaker Hub Evaluator, continue to 2.2.",{"type":50,"tag":69,"props":680,"children":681},{},[682],{"type":56,"value":683},"If user says they do not have one → Continue to 2.2",{"type":50,"tag":322,"props":685,"children":687},{"id":686},"_22-generate-reward-function-from-template",[688],{"type":56,"value":689},"2.2 Generate Reward Function From Template",{"type":50,"tag":65,"props":691,"children":692},{},[693],{"type":50,"tag":69,"props":694,"children":695},{},[696,698,704],{"type":56,"value":697},"Follow workflow in ",{"type":50,"tag":75,"props":699,"children":701},{"className":700},[],[702],{"type":56,"value":703},"references\u002Frlvr_reward_function.md",{"type":56,"value":705}," section \"Helping Users Create Custom Reward Functions\"",{"type":50,"tag":322,"props":707,"children":709},{"id":708},"_23-set-custom_reward_function-value",[710],{"type":56,"value":711},"2.3 Set CUSTOM_REWARD_FUNCTION value",{"type":50,"tag":65,"props":713,"children":714},{},[715],{"type":50,"tag":69,"props":716,"children":717},{},[718,720,726,728,734],{"type":56,"value":719},"Set the value for ",{"type":50,"tag":75,"props":721,"children":723},{"className":722},[],[724],{"type":56,"value":725},"CUSTOM_REWARD_FUNCTION",{"type":56,"value":727}," in the Notebook with the ARN of the reward function (either given directly by the user, or from the function generation code as ",{"type":50,"tag":75,"props":729,"children":731},{"className":730},[],[732],{"type":56,"value":733},"evaluator.arn",{"type":56,"value":735},").",{"type":50,"tag":219,"props":737,"children":739},{"id":738},"_3-rlaif-for-rlaif-only-skip-this-section-if-technique-is-not-rlaif",[740],{"type":56,"value":741},"3. RLAIF (for RLAIF only, skip this section if technique is not RLAIF)",{"type":50,"tag":59,"props":743,"children":744},{},[745,746,752],{"type":56,"value":371},{"type":50,"tag":75,"props":747,"children":749},{"className":748},[],[750],{"type":56,"value":751},"references\u002Frlaif_guide.md",{"type":56,"value":753}," and follow its instructions.",{"type":50,"tag":219,"props":755,"children":757},{"id":756},"_4-eula-review-and-acceptance",[758],{"type":56,"value":759},"4. EULA review and acceptance",{"type":50,"tag":65,"props":761,"children":762},{},[763,768,779],{"type":50,"tag":69,"props":764,"children":765},{},[766],{"type":56,"value":767},"Look up the official license link for the selected base model from references\u002Feula_links.md",{"type":50,"tag":69,"props":769,"children":770},{},[771,773,777],{"type":56,"value":772},"Display the license to the user following the phrasing in references\u002Feula_links.md. For OSS models: \"This model is licensed under ",{"type":50,"tag":161,"props":774,"children":776},{"license":775},"",[],{"type":56,"value":778},". Please review the license terms here: {URL}.\" For Nova models: \"This model is subject to the AWS Service Terms: {URL}.\"",{"type":50,"tag":69,"props":780,"children":781},{},[782,784,789,791],{"type":56,"value":783},"Check if the selected base model is a Meta\u002FLlama model (model ID starts with ",{"type":50,"tag":75,"props":785,"children":787},{"className":786},[],[788],{"type":56,"value":498},{"type":56,"value":790},")\n",{"type":50,"tag":84,"props":792,"children":793},{},[794,827],{"type":50,"tag":69,"props":795,"children":796},{},[797,802,804,810,812,818,820,825],{"type":50,"tag":161,"props":798,"children":799},{},[800],{"type":56,"value":801},"If Meta\u002FLlama",{"type":56,"value":803},": Tell the user they must read and agree to the EULA before using this model. Ask: \"Do you accept the license terms? (yes\u002Fno)\". If the user confirms, set ",{"type":50,"tag":75,"props":805,"children":807},{"className":806},[],[808],{"type":56,"value":809},"ACCEPT_EULA = True",{"type":56,"value":811}," and uncomment ",{"type":50,"tag":75,"props":813,"children":815},{"className":814},[],[816],{"type":56,"value":817},"accept_eula=ACCEPT_EULA",{"type":56,"value":819}," in the generated notebook. If the user declines, leave ",{"type":50,"tag":75,"props":821,"children":823},{"className":822},[],[824],{"type":56,"value":514},{"type":56,"value":826}," and warn that training will fail without acceptance.",{"type":50,"tag":69,"props":828,"children":829},{},[830,835,837,843,845,851],{"type":50,"tag":161,"props":831,"children":832},{},[833],{"type":56,"value":834},"If non-Meta",{"type":56,"value":836},": Inform the user of the license for their awareness. No code-level action needed — the ",{"type":50,"tag":75,"props":838,"children":840},{"className":839},[],[841],{"type":56,"value":842},"ACCEPT_EULA",{"type":56,"value":844}," variable and ",{"type":50,"tag":75,"props":846,"children":848},{"className":847},[],[849],{"type":56,"value":850},"accept_eula",{"type":56,"value":852}," parameter should already be omitted from the notebook (see Step 1.3).",{"type":50,"tag":219,"props":854,"children":856},{"id":855},"_5-post-generation",[857],{"type":56,"value":858},"5. Post-Generation",{"type":50,"tag":59,"props":860,"children":861},{},[862],{"type":56,"value":863},"After generating the code, offer to run it. Training can take hours depending on your dataset and model.",{"type":50,"tag":59,"props":865,"children":866},{},[867,872,874,879],{"type":50,"tag":161,"props":868,"children":869},{},[870],{"type":56,"value":871},"Notebook mode:",{"type":56,"value":873}," If ",{"type":50,"tag":75,"props":875,"children":877},{"className":876},[],[878],{"type":56,"value":292},{"type":56,"value":880}," is available, offer to run the cells. Otherwise tell the user to run cells themselves.",{"type":50,"tag":59,"props":882,"children":883},{},[884,889],{"type":50,"tag":161,"props":885,"children":886},{},[887],{"type":56,"value":888},"Script mode:",{"type":56,"value":890}," Present the user with options:",{"type":50,"tag":892,"props":893,"children":894},"blockquote",{},[895,900],{"type":50,"tag":59,"props":896,"children":897},{},[898],{"type":56,"value":899},"\"Would you like me to:",{"type":50,"tag":65,"props":901,"children":902},{},[903,914,919],{"type":50,"tag":69,"props":904,"children":905},{},[906,908],{"type":56,"value":907},"Leave it to you — run with ",{"type":50,"tag":75,"props":909,"children":911},{"className":910},[],[912],{"type":56,"value":913},"python scripts\u002F[script_name]",{"type":50,"tag":69,"props":915,"children":916},{},[917],{"type":56,"value":918},"Run it and wait until it's done",{"type":50,"tag":69,"props":920,"children":921},{},[922],{"type":56,"value":923},"Start it but don't wait — we can check status later\"",{"type":50,"tag":84,"props":925,"children":926},{},[927,937,955],{"type":50,"tag":69,"props":928,"children":929},{},[930,935],{"type":50,"tag":161,"props":931,"children":932},{},[933],{"type":56,"value":934},"Option 1:",{"type":56,"value":936}," Done. Wait for user to come back.",{"type":50,"tag":69,"props":938,"children":939},{},[940,945,947,953],{"type":50,"tag":161,"props":941,"children":942},{},[943],{"type":56,"value":944},"Option 2:",{"type":56,"value":946}," Execute the script as-is. ",{"type":50,"tag":75,"props":948,"children":950},{"className":949},[],[951],{"type":56,"value":952},"trainer.train(wait=True)",{"type":56,"value":954}," blocks until complete. Report final status.",{"type":50,"tag":69,"props":956,"children":957},{},[958,963,965,971,973,979],{"type":50,"tag":161,"props":959,"children":960},{},[961],{"type":56,"value":962},"Option 3:",{"type":56,"value":964}," Change ",{"type":50,"tag":75,"props":966,"children":968},{"className":967},[],[969],{"type":56,"value":970},"wait=True",{"type":56,"value":972}," to ",{"type":50,"tag":75,"props":974,"children":976},{"className":975},[],[977],{"type":56,"value":978},"wait=False",{"type":56,"value":980}," in the script, execute, report the training job name.",{"type":50,"tag":59,"props":982,"children":983},{},[984],{"type":50,"tag":161,"props":985,"children":986},{},[987],{"type":56,"value":988},"Checking status:",{"type":50,"tag":84,"props":990,"children":991},{},[992,1024],{"type":50,"tag":69,"props":993,"children":994},{},[995,1001,1003,1009,1011,1017,1018],{"type":50,"tag":75,"props":996,"children":998},{"className":997},[],[999],{"type":56,"value":1000},"describe-training-job --training-job-name NAME",{"type":56,"value":1002}," → ",{"type":50,"tag":75,"props":1004,"children":1006},{"className":1005},[],[1007],{"type":56,"value":1008},"TrainingJobStatus",{"type":56,"value":1010},", ",{"type":50,"tag":75,"props":1012,"children":1014},{"className":1013},[],[1015],{"type":56,"value":1016},"FailureReason",{"type":56,"value":1010},{"type":50,"tag":75,"props":1019,"children":1021},{"className":1020},[],[1022],{"type":56,"value":1023},"SecondaryStatusTransitions",{"type":50,"tag":69,"props":1025,"children":1026},{},[1027,1029],{"type":56,"value":1028},"For model package ARN after completion: ",{"type":50,"tag":75,"props":1030,"children":1032},{"className":1031},[],[1033],{"type":56,"value":1034},"list-model-packages --model-package-group-name GROUP_NAME --sort-by CreationTime --sort-order Descending --max-results 1",{"type":50,"tag":59,"props":1036,"children":1037},{},[1038],{"type":50,"tag":161,"props":1039,"children":1040},{},[1041],{"type":56,"value":1042},"Showing results after completion:",{"type":50,"tag":84,"props":1044,"children":1045},{},[1046,1059],{"type":50,"tag":69,"props":1047,"children":1048},{},[1049,1051,1057],{"type":56,"value":1050},"Use ",{"type":50,"tag":75,"props":1052,"children":1054},{"className":1053},[],[1055],{"type":56,"value":1056},"scripts\u002Fmlflow_reference.py",{"type":56,"value":1058}," as the pattern to query MLflow metrics",{"type":50,"tag":69,"props":1060,"children":1061},{},[1062],{"type":56,"value":1063},"Present loss by epoch as a text table (total_loss, val_eval_total_loss for SFT; rewards\u002Fmargins for DPO; critic\u002Frewards\u002Fmean for RLVR)",{"type":50,"tag":59,"props":1065,"children":1066},{},[1067],{"type":50,"tag":161,"props":1068,"children":1069},{},[1070],{"type":56,"value":1071},"CRITICAL:",{"type":50,"tag":84,"props":1073,"children":1074},{},[1075,1080],{"type":50,"tag":69,"props":1076,"children":1077},{},[1078],{"type":56,"value":1079},"DON'T suggest moving to next steps before training completes",{"type":50,"tag":69,"props":1081,"children":1082},{},[1083],{"type":56,"value":1084},"DON'T elaborate on the next steps unless the user specifically asks you about them.",{"type":50,"tag":219,"props":1086,"children":1088},{"id":1087},"_6-continuous-customization",[1089],{"type":56,"value":1090},"6. Continuous Customization",{"type":50,"tag":59,"props":1092,"children":1093},{},[1094],{"type":56,"value":1095},"If the user wants to finetune a model they had already customized, follow the instructions in references\u002Fcontinuous_customization.md",{"type":50,"tag":209,"props":1097,"children":1098},{},[],{"type":50,"tag":51,"props":1100,"children":1102},{"id":1101},"references",[1103],{"type":56,"value":1104},"References",{"type":50,"tag":84,"props":1106,"children":1107},{},[1108,1119,1130,1141,1151,1161,1171,1182,1193,1204],{"type":50,"tag":69,"props":1109,"children":1110},{},[1111,1117],{"type":50,"tag":75,"props":1112,"children":1114},{"className":1113},[],[1115],{"type":56,"value":1116},"rlvr_reward_function.md",{"type":56,"value":1118}," - Lambda reward function creation guide (RLVR only)",{"type":50,"tag":69,"props":1120,"children":1121},{},[1122,1128],{"type":50,"tag":75,"props":1123,"children":1125},{"className":1124},[],[1126],{"type":56,"value":1127},"templates\u002Frlvr_reward_function_source_template.py",{"type":56,"value":1129}," - Lambda reward function source template for open-weights models (RLVR only)",{"type":50,"tag":69,"props":1131,"children":1132},{},[1133,1139],{"type":50,"tag":75,"props":1134,"children":1136},{"className":1135},[],[1137],{"type":56,"value":1138},"templates\u002Fnova_rlvr_reward_function_source_template.py",{"type":56,"value":1140}," - Lambda reward function source template for Nova 2.0 Lite (RLVR only)",{"type":50,"tag":69,"props":1142,"children":1143},{},[1144,1149],{"type":50,"tag":75,"props":1145,"children":1147},{"className":1146},[],[1148],{"type":56,"value":393},{"type":56,"value":1150}," - Complete notebook template for Supervised Fine-Tuning (OSS path)",{"type":50,"tag":69,"props":1152,"children":1153},{},[1154,1159],{"type":50,"tag":75,"props":1155,"children":1157},{"className":1156},[],[1158],{"type":56,"value":404},{"type":56,"value":1160}," - Complete notebook template for Direct Preference Optimization (OSS path)",{"type":50,"tag":69,"props":1162,"children":1163},{},[1164,1169],{"type":50,"tag":75,"props":1165,"children":1167},{"className":1166},[],[1168],{"type":56,"value":415},{"type":56,"value":1170}," - Complete notebook template for Reinforcement Learning from Verifiable Rewards (OSS path)",{"type":50,"tag":69,"props":1172,"children":1173},{},[1174,1180],{"type":50,"tag":75,"props":1175,"children":1177},{"className":1176},[],[1178],{"type":56,"value":1179},"references\u002Fcontinuous_customization.md",{"type":56,"value":1181}," - Instructions on fine-tuning an already fine-tuned model.",{"type":50,"tag":69,"props":1183,"children":1184},{},[1185,1191],{"type":50,"tag":75,"props":1186,"children":1188},{"className":1187},[],[1189],{"type":56,"value":1190},"rlaif_guide.md",{"type":56,"value":1192}," - instructions on RLAIF finetuning options",{"type":50,"tag":69,"props":1194,"children":1195},{},[1196,1202],{"type":50,"tag":75,"props":1197,"children":1199},{"className":1198},[],[1200],{"type":56,"value":1201},"rlaif_builtin.py",{"type":56,"value":1203}," - Code template for RLAIF with built-in judge prompt",{"type":50,"tag":69,"props":1205,"children":1206},{},[1207,1213],{"type":50,"tag":75,"props":1208,"children":1210},{"className":1209},[],[1211],{"type":56,"value":1212},"rlaif_custom_prompt.py",{"type":56,"value":1214}," - Code template for RLAIF with custom judge prompt",{"items":1216,"total":1321},[1217,1234,1254,1266,1283,1296,1306],{"slug":1218,"name":1218,"fn":1219,"description":1220,"org":1221,"tags":1222,"stars":26,"repoUrl":27,"updatedAt":1233},"amazon-location-service","integrate Amazon Location Service maps","Integrates Amazon Location Service APIs for AWS applications. Use this skill when users want to add maps (interactive MapLibre or static images); geocode addresses to coordinates or reverse geocode coordinates to addresses; calculate routes, travel times, or service areas; find places and businesses through text search, nearby search, or autocomplete suggestions; retrieve detailed place information including hours, contacts, and addresses; monitor geographical boundaries with geofences; or track device locations. Covers authentication, SDK integration, and all Amazon Location Service capabilities.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1223,1226,1227,1230],{"name":1224,"slug":1225,"type":16},"API Development","api-development",{"name":24,"slug":25,"type":16},{"name":1228,"slug":1229,"type":16},"Maps","maps",{"name":1231,"slug":1232,"type":16},"Navigation","navigation","2026-07-12T08:39:49.88311",{"slug":1235,"name":1235,"fn":1236,"description":1237,"org":1238,"tags":1239,"stars":26,"repoUrl":27,"updatedAt":1253},"amplify-workflow","build and deploy apps with AWS Amplify","Build and deploy full-stack web and mobile apps with AWS Amplify Gen2 (TypeScript code-first). Covers auth (Cognito), data (AppSync\u002FDynamoDB including schema modeling, enum types, relationships, authorization rules), storage (S3), functions, APIs, and AI (Amplify AI Kit with Bedrock). Supports React, Next.js, Vue, Angular, React Native, Flutter, Swift, and Android. Always use this skill for Amplify Gen2 topics — even for questions you think you know — it contains validated, version-specific patterns that prevent common mistakes. TRIGGER when: user mentions Amplify Gen2; project has amplify\u002F directory or amplify_outputs; code imports @aws-amplify packages; user asks about defineBackend, defineAuth, defineData, defineStorage, or npx ampx. SKIP: Amplify Gen1 (amplify CLI v6), standalone SAM\u002FCDK without Amplify (use aws-serverless), direct Bedrock without Amplify AI Kit (use bedrock).",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1240,1243,1244,1247,1250],{"name":1241,"slug":1242,"type":16},"Auth","auth",{"name":24,"slug":25,"type":16},{"name":1245,"slug":1246,"type":16},"Database","database",{"name":1248,"slug":1249,"type":16},"Deployment","deployment",{"name":1251,"slug":1252,"type":16},"TypeScript","typescript","2026-07-12T08:39:43.500162",{"slug":1255,"name":1255,"fn":1256,"description":1257,"org":1258,"tags":1259,"stars":26,"repoUrl":27,"updatedAt":1265},"api-gateway","build and manage Amazon API Gateway APIs","Build, manage, and operate APIs with Amazon API Gateway (REST, HTTP, and WebSocket). Triggers on phrases like: API Gateway, REST API, HTTP API, WebSocket API, custom domain, Lambda authorizer, usage plan, throttling, CORS, VPC link, private API. Also covers troubleshooting API Gateway errors (4xx, 5xx, timeout, CORS failures) and IaC templates containing API Gateway resources. For general REST API design unrelated to AWS, do not trigger.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1260,1261,1262],{"name":1224,"slug":1225,"type":16},{"name":24,"slug":25,"type":16},{"name":1263,"slug":1264,"type":16},"REST API","rest-api","2026-07-12T08:39:00.149339",{"slug":1267,"name":1267,"fn":1268,"description":1269,"org":1270,"tags":1271,"stars":26,"repoUrl":27,"updatedAt":1282},"aws-architecture-diagram","generate AWS architecture diagrams","Generate validated AWS architecture diagrams as draw.io XML using official AWS4 icon libraries. Use this skill whenever the user wants to create, generate, or design AWS architecture diagrams, cloud infrastructure diagrams, or system design visuals. Also triggers for requests to visualize existing infrastructure from CloudFormation, CDK, or Terraform code. Supports two modes: analyze an existing codebase to auto-generate diagrams, or brainstorm interactively from scratch. Exports .drawio files with optional PNG\u002FSVG\u002FPDF export via draw.io desktop CLI.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1272,1275,1276,1279],{"name":1273,"slug":1274,"type":16},"Architecture","architecture",{"name":24,"slug":25,"type":16},{"name":1277,"slug":1278,"type":16},"Design","design",{"name":1280,"slug":1281,"type":16},"Diagrams","diagrams","2026-07-12T08:37:11.012278",{"slug":1284,"name":1284,"fn":1285,"description":1286,"org":1287,"tags":1288,"stars":26,"repoUrl":27,"updatedAt":1295},"aws-lambda","build and deploy AWS Lambda functions","Design, build, deploy, test, and debug serverless applications with AWS Lambda. Triggers on phrases like: Lambda function, event source, serverless application, API Gateway, EventBridge, Step Functions, serverless API, event-driven architecture, Lambda trigger. For deploying non-serverless apps to AWS, use deploy-on-aws plugin instead.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1289,1290,1291,1292],{"name":1224,"slug":1225,"type":16},{"name":24,"slug":25,"type":16},{"name":1248,"slug":1249,"type":16},{"name":1293,"slug":1294,"type":16},"Serverless","serverless","2026-07-12T08:38:58.598492",{"slug":1297,"name":1297,"fn":1298,"description":1299,"org":1300,"tags":1301,"stars":26,"repoUrl":27,"updatedAt":1305},"aws-lambda-durable-functions","build resilient AWS Lambda durable functions","Build resilient, long-running, multi-step applications with AWS Lambda durable functions with automatic state persistence, retry logic, and orchestration for long-running executions. Covers the critical replay model, step operations, wait\u002Fcallback patterns, error handling with saga pattern, testing with LocalDurableTestRunner. Triggers on phrases like: lambda durable functions, workflow orchestration, state machines, retry\u002Fcheckpoint patterns, long-running stateful Lambda functions, saga pattern, human-in-the-loop callbacks, and reliable serverless applications.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1302,1303,1304],{"name":1273,"slug":1274,"type":16},{"name":24,"slug":25,"type":16},{"name":1293,"slug":1294,"type":16},"2026-07-12T08:39:05.546173",{"slug":1307,"name":1307,"fn":1308,"description":1309,"org":1310,"tags":1311,"stars":26,"repoUrl":27,"updatedAt":1320},"aws-lambda-managed-instances","configure AWS Lambda Managed Instances","Evaluate, configure, and migrate workloads to AWS Lambda Managed Instances (LMI). Triggers on: Lambda Managed Instances, LMI, capacity provider, multi-concurrency Lambda, dedicated instance Lambda, EC2-backed Lambda, cold start elimination, Graviton Lambda, instance type for Lambda, scheduled scaling for LMI, Lambda cost optimization with Reserved Instances or Savings Plans. Also trigger when users describe high-volume predictable workloads seeking cost savings, want to scale LMI capacity on a schedule, or compare Lambda vs EC2 for steady-state traffic. For standard Lambda without LMI, use the aws-lambda skill instead.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1312,1313,1314,1317],{"name":24,"slug":25,"type":16},{"name":1248,"slug":1249,"type":16},{"name":1315,"slug":1316,"type":16},"Infrastructure","infrastructure",{"name":1318,"slug":1319,"type":16},"Performance","performance","2026-07-12T08:39:07.007071",33,{"items":1323,"total":1497},[1324,1343,1360,1370,1383,1396,1406,1416,1437,1452,1467,1482],{"slug":1325,"name":1325,"fn":1326,"description":1327,"org":1328,"tags":1329,"stars":1340,"repoUrl":1341,"updatedAt":1342},"agentcore-investigation","investigate Bedrock AgentCore runtime sessions","Investigate Bedrock AgentCore runtime sessions via CloudWatch Logs Insights — resolve session\u002Ftrace IDs, query OTEL spans, filter noise, build timelines. Use when debugging AgentCore agent sessions, tracing tool calls, or analyzing latency.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1330,1331,1334,1337],{"name":24,"slug":25,"type":16},{"name":1332,"slug":1333,"type":16},"Debugging","debugging",{"name":1335,"slug":1336,"type":16},"Logs","logs",{"name":1338,"slug":1339,"type":16},"Observability","observability",9427,"https:\u002F\u002Fgithub.com\u002Fawslabs\u002Fmcp","2026-07-12T08:37:22.601527",{"slug":1344,"name":1345,"fn":1346,"description":1347,"org":1348,"tags":1349,"stars":1340,"repoUrl":1341,"updatedAt":1359},"amazon-aurora-dsql","amazon aurora dsql","build applications with Aurora DSQL","Build with Aurora DSQL — manage schemas, execute queries, handle migrations, diagnose query plans, load data, and develop applications with a serverless, distributed SQL database. Covers IAM auth, multi-tenant patterns, MySQL-to-DSQL and PostgreSQL-to-DSQL schema conversion, FK replacement code generation, OCC retry patterns, ORM migration (Django\u002FHibernate\u002FRails), DDL operations, query plan explainability, SQL compatibility validation, and bulk data loading. Triggers on phrases like: DSQL, Aurora DSQL, create DSQL table, DSQL schema, migrate to DSQL, distributed SQL database, serverless PostgreSQL-compatible database, DSQL query plan, DSQL EXPLAIN ANALYZE, why is my DSQL query slow, DSQL foreign key, DSQL OCC retry, DSQL multi-region, load into DSQL, load CSV into DSQL, bulk load DSQL, aurora-dsql-loader.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1350,1353,1354,1355,1356],{"name":1351,"slug":1352,"type":16},"Aurora","aurora",{"name":24,"slug":25,"type":16},{"name":1245,"slug":1246,"type":16},{"name":1293,"slug":1294,"type":16},{"name":1357,"slug":1358,"type":16},"SQL","sql","2026-07-12T08:36:45.053393",{"slug":1361,"name":1362,"fn":1346,"description":1347,"org":1363,"tags":1364,"stars":1340,"repoUrl":1341,"updatedAt":1369},"aurora-dsql","aurora dsql",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1365,1366,1367,1368],{"name":24,"slug":25,"type":16},{"name":1245,"slug":1246,"type":16},{"name":1293,"slug":1294,"type":16},{"name":1357,"slug":1358,"type":16},"2026-07-12T08:36:42.694299",{"slug":1371,"name":1372,"fn":1346,"description":1347,"org":1373,"tags":1374,"stars":1340,"repoUrl":1341,"updatedAt":1382},"aws-dsql","aws dsql",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1375,1376,1377,1380,1381],{"name":24,"slug":25,"type":16},{"name":1245,"slug":1246,"type":16},{"name":1378,"slug":1379,"type":16},"Migration","migration",{"name":1293,"slug":1294,"type":16},{"name":1357,"slug":1358,"type":16},"2026-07-12T08:36:38.584057",{"slug":1384,"name":1385,"fn":1346,"description":1347,"org":1386,"tags":1387,"stars":1340,"repoUrl":1341,"updatedAt":1395},"distributed-postgres","distributed postgres",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1388,1389,1390,1393,1394],{"name":24,"slug":25,"type":16},{"name":1245,"slug":1246,"type":16},{"name":1391,"slug":1392,"type":16},"PostgreSQL","postgresql",{"name":1293,"slug":1294,"type":16},{"name":1357,"slug":1358,"type":16},"2026-07-12T08:36:46.530743",{"slug":1397,"name":1398,"fn":1346,"description":1347,"org":1399,"tags":1400,"stars":1340,"repoUrl":1341,"updatedAt":1405},"distributed-sql","distributed sql",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1401,1402,1403,1404],{"name":24,"slug":25,"type":16},{"name":1245,"slug":1246,"type":16},{"name":1293,"slug":1294,"type":16},{"name":1357,"slug":1358,"type":16},"2026-07-12T08:36:48.104182",{"slug":1407,"name":1407,"fn":1346,"description":1347,"org":1408,"tags":1409,"stars":1340,"repoUrl":1341,"updatedAt":1415},"dsql",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1410,1411,1412,1413,1414],{"name":24,"slug":25,"type":16},{"name":1245,"slug":1246,"type":16},{"name":1378,"slug":1379,"type":16},{"name":1293,"slug":1294,"type":16},{"name":1357,"slug":1358,"type":16},"2026-07-12T08:36:36.374512",{"slug":1417,"name":1417,"fn":1418,"description":1419,"org":1420,"tags":1421,"stars":1434,"repoUrl":1435,"updatedAt":1436},"cost-efficiency-analyzer","analyze cost efficiency and expenses","Analyzes cost structure, cost efficiency, and expense management from P&L data. Use when the user asks about costs, expenses, COGS, operating expenses, cost ratios, cost control, spending efficiency, margin compression from cost side, or wants to understand where money is going. Also use for \"are we spending too much\", \"cost breakdown\", \"expense analysis\", or \"how efficient are our operations\". NOT for revenue or top-line analysis.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1422,1425,1428,1431],{"name":1423,"slug":1424,"type":16},"Accounting","accounting",{"name":1426,"slug":1427,"type":16},"Analytics","analytics",{"name":1429,"slug":1430,"type":16},"Cost Optimization","cost-optimization",{"name":1432,"slug":1433,"type":16},"Finance","finance",3176,"https:\u002F\u002Fgithub.com\u002Fawslabs\u002Fagentcore-samples","2026-07-12T08:40:03.29555",{"slug":1438,"name":1438,"fn":1439,"description":1440,"org":1441,"tags":1442,"stars":1434,"repoUrl":1435,"updatedAt":1451},"executive-financial-briefing","generate executive financial briefings","Generates a concise executive-level financial briefing or summary suitable for a CEO, CFO, or board presentation. Use when the user asks for a summary, briefing, executive summary, board update, financial overview, financial health check, or \"how is the business doing\". Covers the full P&L picture in one page. Also use for \"give me the highlights\", \"what do I need to know\", or \"quick financial update\".",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1443,1444,1445,1448],{"name":24,"slug":25,"type":16},{"name":1432,"slug":1433,"type":16},{"name":1446,"slug":1447,"type":16},"Management","management",{"name":1449,"slug":1450,"type":16},"Reporting","reporting","2026-07-12T08:40:02.066471",{"slug":1453,"name":1453,"fn":1454,"description":1455,"org":1456,"tags":1457,"stars":1434,"repoUrl":1435,"updatedAt":1466},"multi-quarter-trend-analysis","analyze multi-quarter financial trends","Analyzes financial trends across multiple quarters by comparing P&L metrics over time. Use when the user wants to see trends, patterns, trajectories, or directional movement across 3 or more quarters. Also use for \"how are we trending\", \"show me the trend\", \"track performance over time\", \"quarter over quarter comparison across all quarters\", or any multi-period longitudinal analysis.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1458,1459,1460,1463],{"name":1426,"slug":1427,"type":16},{"name":1432,"slug":1433,"type":16},{"name":1461,"slug":1462,"type":16},"Financial Statements","financial-statements",{"name":1464,"slug":1465,"type":16},"Variance Analysis","variance-analysis","2026-07-12T08:40:00.79141",{"slug":1468,"name":1468,"fn":1469,"description":1470,"org":1471,"tags":1472,"stars":1434,"repoUrl":1435,"updatedAt":1481},"pdf","process and manipulate PDF documents","Use this skill whenever the user wants to do anything with PDF files. This includes reading or extracting text\u002Ftables from PDFs, combining or merging multiple PDFs into one, splitting PDFs apart, rotating pages, adding watermarks, creating new PDFs, filling PDF forms, encrypting\u002Fdecrypting PDFs, extracting images, and OCR on scanned PDFs to make them searchable. If the user mentions a .pdf file or asks to produce one, use this skill.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1473,1476,1479],{"name":1474,"slug":1475,"type":16},"Automation","automation",{"name":1477,"slug":1478,"type":16},"Documents","documents",{"name":1480,"slug":1468,"type":16},"PDF","2026-07-12T08:41:44.135656",{"slug":1483,"name":1483,"fn":1484,"description":1485,"org":1486,"tags":1487,"stars":1434,"repoUrl":1435,"updatedAt":1496},"quarterly-kpi-calculator","calculate quarterly financial KPIs","Calculates quarterly financial KPIs from P&L data. P&L figures can be provided directly by the user or fetched from the financial data MCP server. Use when the user wants KPI calculations such as Gross Margin %, EBITDA Margin %, Operating Expense Ratio, or Revenue Growth % QoQ. Also use for quarterly performance review, P&L analysis, or interpreting financial ratios against benchmarks.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1488,1489,1492,1493],{"name":1423,"slug":1424,"type":16},{"name":1490,"slug":1491,"type":16},"Data Analysis","data-analysis",{"name":1432,"slug":1433,"type":16},{"name":1494,"slug":1495,"type":16},"KPI","kpi","2026-07-12T08:39:59.54971",150]