[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill-aws-labs-dataset-transformation":3,"mdc-8rwqyd-key":43,"related-org-aws-labs-dataset-transformation":1206,"related-repo-aws-labs-dataset-transformation":1387},{"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":38,"sourceUrl":41,"mdContent":42},"dataset-transformation","transform datasets for ML model training","Generates code that transforms datasets between ML schemas for model training or evaluation. Use when the user says \"transform\", \"convert\", \"reformat\", \"change the format\", or when a dataset's schema needs to change to match the target format — always use this skill for format changes rather than writing inline transformation code. Supports OpenAI chat, SageMaker SFT\u002FDPO\u002FRLVR\u002FRLAIF, HuggingFace preference, Bedrock Nova, VERL, and custom JSONL formats from local files or S3.",{"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},"Datasets","datasets","tag",{"name":18,"slug":19,"type":16},"Data Engineering","data-engineering",{"name":21,"slug":22,"type":16},"Data Pipeline","data-pipeline",{"name":24,"slug":25,"type":16},"Python","python",831,"https:\u002F\u002Fgithub.com\u002Fawslabs\u002Fagent-plugins","2026-07-12T08:39:27.993604",null,127,[32,33,34,35,36,37],"agent-plugins","agent-skills","agents","aws","coding-agent-skills","coding-agents",{"repoUrl":27,"stars":26,"forks":30,"topics":39,"description":40},[32,33,34,35,36,37],"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\u002Fdataset-transformation","---\nname: dataset-transformation\ndescription: Generates code that transforms datasets between ML schemas for model training or evaluation. Use when the user says \"transform\", \"convert\", \"reformat\", \"change the format\", or when a dataset's schema needs to change to match the target format — always use this skill for format changes rather than writing inline transformation code. Supports OpenAI chat, SageMaker SFT\u002FDPO\u002FRLVR\u002FRLAIF, HuggingFace preference, Bedrock Nova, VERL, and custom JSONL formats from local files or S3.\nmetadata:\n  version: \"1.0.0\"\n---\n\n# Dataset Transformation Agent\n\nTransforms a data set provided by the user into their desired format.\n\n## When to Use\n\n- User needs to generate code for transforming datasets for SageMaker model training or model evaluation.\n- A dataset requires processing, cleaning, or formatting before training or evaluation.\n- Workflow requires a formal review and approval cycle before execution.\n\n## Prerequisites\n\n- The SDK environment has been verified (SDK version, region, execution role). If not done, activate the `sdk-getting-started` skill first.\n\n## Principles\n\n1. **One thing at a time.** Each response advances exactly one decision. Never combine multiple questions or recommendations in a single turn.\n2. **Confirm before proceeding.** Wait for the user to agree before moving to the next step. You are a guide, not a runaway train.\n3. **Don't read files until you need them.** Only read reference files when you've reached the workflow step that requires them and the user has confirmed the direction. Never read ahead.\n4. **No narration.** Don't explain what you're about to do or what you just did. Share outcomes and ask questions. Keep responses short and focused.\n5. **No repetition.** If you said something before a tool call, don't repeat it after. Only share new information.\n6. **Do not deviate from the Workflow.** The steps listed in the workflow should be followed exactly as described. Progress from Step 1 to Step 11 to complete the task. Do not deviate from the workflow!\n7. **Always end with a question.** Whenever you pause for user input, acknowledgment, or feedback, your response must end with a question. Never leave the user with a statement and expect them to know they need to respond.\n8. **Default output format is JSONL.** Unless the user explicitly requests a different file format, the transformed dataset should be written as `.jsonl` (JSON Lines — one JSON object per line).\n\n## Known Dataset Formats Reference\n\nThis skill supports two transformation purposes — **training data** and **evaluation data** — each with its own format resolution path. The purpose is determined in Step 1 of the workflow.\n\n### Training Data Formats\n\nResolve the target format using the reference file ..\u002Fdataset-evaluation\u002Freferences\u002Fstrategy_data_requirements.md. When the transformation is for **model training**, the required format depends on both the **model type** (Open Weights like Llama\u002FQwen vs Nova) and the **finetuning technique** (SFT, DPO, RLVR, RLAIF) — make sure to match on both dimensions. If either the model type or technique is not yet known, ask the user before resolving the format.\n\n### Evaluation Data Formats\n\nWhen the transformation is for **model evaluation**, resolve the target format using this order:\n\n1. Try fetching the live documentation at https:\u002F\u002Fdocs.aws.amazon.com\u002Fsagemaker\u002Flatest\u002Fdg\u002Fmodel-customize-evaluation-dataset-formats.html to get the latest evaluation dataset schema definitions.\n2. **If the fetch fails** (e.g., no internet access, VPC environment), fall back to the offline copy at `references\u002Fsagemaker_dataset_formats.md`. Inform the user that the format schemas are from an offline copy and may be outdated.\n\nUse whichever source you successfully access as the source of truth for the target format. Do not rely on memorized schemas.\n\n## Workflow\n\n### Step 1: Determine transformation purpose\n\nYour first response should determine whether this transformation is for **model training** or **model evaluation**. If the context already makes this clear (e.g., the user said \"I need to prep my training data\" or \"I need to format my eval dataset\"), confirm your understanding and move on. Otherwise, ask:\n\n> \"Is this dataset transformation for model training or model evaluation? This helps me look up the right target format for you.\"\n\n- **Training** → format resolution will use the local training data requirements reference (model type + finetuning technique dependent).\n- **Evaluation** → format resolution will use the live AWS documentation (with offline fallback).\n\nRemember this choice — it determines how the target format is resolved in Step 3.\n\n⏸ Wait for user.\n\n### Step 2: Set expectations\n\nAcknowledge the user's request and state what this skill can do:\n\n> \"I can help you transform your dataset's format! Here's my plan: I will first need to understand the format of your dataset and the transformation requirements. Once I have that, I will generate a dataset transformation function that we can refine together. After the dataset transformation function is refined to your liking, I will perform the transformation task and upload it to your desired location! Does this sound good?\"\n\n⏸ Wait for user.\n\n### Step 3: Understand the dataset transformation task\n\nFor this step, you need to know: **what dataset format the user would like to transform their dataset from and what dataset format they would like to transform it in to.**\nIf you know this already, skip this step. If not, ask the user:\n\n> \"What's the dataset format you would like to transform it into?\"\n\nResolve the target format based on the purpose determined in Step 1:\n\n- **If training data**: Ask the user for the finetuning technique (SFT, DPO, RLVR, RLAIF) and model type (Open Weights like Llama\u002FQwen vs Nova) if not already known. Then look up the required format from the \"Training Data Formats\" section in the Known Dataset Formats Reference above.\n- **If evaluation data**: If the user mentions a well-known format name (e.g., \"OpenAI format\", \"SageMaker format\"), fetch the schema from the live documentation as described in the \"Evaluation Data Formats\" section above. If a well-known format is fetched, confirm with the user:\n\n> \"I've found a SageMaker dataset format: {sagemaker-dataset-format-name} with schema: {sagemaker-dataset-format-schema}. Is this what you were referring to?\"\n\nIf the user describes a custom format not listed in the reference doc, ask them to provide a sample record of the desired output format.\n\n⏸ Wait for user.\n\n### Step 4: Get the dataset from the user\n\nFor this step, you need: **the location of the user's dataset**.\nIf you know this already, skip this step. If not, ask the user:\n\n> \"Where can I find your dataset? Either a local directory or S3 location works!\"\n\n⏸ Wait for user.\n\n### Step 5: Examine sample data\n\nRead 1–2 sample records from the user's dataset and show them so the user can confirm the source schema. Do not run format detection — that is handled by the planning skill before this skill is invoked.\n\nDo not show a side-by-side mapping to the target format here — the detailed mapping will be handled in Step 7 when generating the transformation function.\n\n⏸ Wait for user.\n\n### Step 6: Get the dataset output location\n\nFor this step, you need: **to understand where to output the transformed dataset to. It could be an S3 URI or local directory**\nIf you already know where the dataset is supposed to be output to, skip this step. If not, ask the user:\n\n> \"Where should I output your transformed dataset to? Either a local directory or S3 location works!\"\n\nIf the user provides a directory (not a full file path), construct the output filename using the pattern `{original_name}_{target_format}.jsonl` (e.g., `gen_qa_100k_openai.jsonl`).\n\n⏸ Wait for user.\n\n### Step 7: Generate and validate the transformation function\n\nFor this step, you need: **to generate a python function that transforms the dataset from the format in Step 5 to the format in Step 3**\n\nRead the reference guide at `references\u002Fdataset_transformation_code.md` and follow its skeleton exactly when generating the transformation function.\n\nThe python function should be in the form of:\n\n```python\ndef transform_dataset(df: pd.DataFrame) -> pd.DataFrame:\n```\n\nThe `\u003Cproject-dir>` is the project directory established by the directory-management skill (e.g., `dpo-to-rlvr-conversion`).\n\nIn notebook mode, add a `%%writefile \u003Cproject-dir>\u002Fscripts\u002Ftransform_fn.py` code cell AND write the file to disk for testing. In script mode, write the file to disk directly.\n\nContinue iterating with the user's feedback — update the code in place on each revision rather than showing code inline.\n\n**If sample data was collected in Step 5**, test the function against the sample records:\n\n1. Generate the transformation function.\n2. Write the sample data to a temporary JSONL file (e.g., `\u002Ftmp\u002Ftest_input.jsonl`), then run:\n   `python3 -c \"import sys; sys.path.insert(0, '\u003Cproject-dir>\u002Fscripts'); from transform_fn import transform_dataset; import pandas as pd; df = pd.read_json('\u002Ftmp\u002Ftest_input.jsonl', lines=True); result = transform_dataset(df); print(result.to_json(orient='records', lines=True))\"`\n3. If the test fails, fix and re-test until it passes.\n4. Show the user the function and transformed sample output for review.\n\n**If no sample data**, present the function for review and refinement.\n\n⏸ Wait for user.\n\n### Step 8: Determine output target\n\nIf no project directory exists, activate the **directory-management** skill to set one up.\n\n⏸ Wait for user.\n\n### Step 9: Generate the execution code\n\n**Before writing the code, read:**\n\n- `references\u002Fcode_output_guide.md` (output format rules)\n- `code_templates\u002Ftransformation.py` (cell structure and skeleton code)\n\nThe template uses `# Cell N: Label` markers — each marker starts a new section. Cell 2 (Transformation Function) is dynamically generated from Step 7; all other cells follow the template skeleton.\n\nGenerate the execution logic following the code output guide.\n\n- In notebook mode, add a `%%writefile \u003Cproject-dir>\u002Fscripts\u002F\u003Cscript_name>.py` code cell AND write the file to disk. In script mode, write the file to disk directly.\n- The script must import `transform_dataset` from `transform_fn`.\n- Replace placeholders with the actual input\u002Foutput paths.\n\nRead the reference guide at `references\u002Fdataset_transformation_code.md` and follow its execution script skeleton exactly.\n\n**If sample data was collected in Step 5**, test the full pipeline:\n\n1. Write the sample records to a temporary JSONL file (e.g., `\u002Ftmp\u002Ftest_input.jsonl`).\n2. Run: `python3 \u003Cproject-dir>\u002Fscripts\u002F\u003Cscript_name> --input \u002Ftmp\u002Ftest_input.jsonl --output \u002Ftmp\u002Ftest_output.jsonl`\n3. If it fails, debug and fix, then re-run until successful.\n4. Show the user the output for review.\n\n**If no sample data**, present the notebook for review and refinement.\n\n⏸ Wait for user.\n\n### Step 10: Determine and confirm execution mode\n\nCheck the size of the input dataset:\n\n- If the dataset is in S3, use the AWS MCP tool `head-object` (S3 service) with the bucket and key to get `ContentLength`.\n- If the dataset is local, check the file size.\n\n**Decision criteria:**\n\n- Dataset \u003C 50 MB → recommend local execution\n- Dataset ≥ 50 MB → recommend SageMaker Processing Job\n\nInform the user of the recommendation and get their approval:\n\nIf local:\n\n> \"Your dataset is {size} MB — since it's under 50 MB, I'd recommend running the transformation locally. Would you like to proceed with local execution, or would you prefer a SageMaker Processing Job instead?\"\n\nIf SageMaker Processing Job:\n\n> \"Your dataset is {size} MB — since it's over 50 MB, I'd recommend running this as a SageMaker Processing Job for better performance. Would you like to proceed with a SageMaker Processing Job, or would you prefer to run it locally instead?\"\n\nDo not execute until the user approves. If the user rejects the recommendation, switch to the alternative and get their explicit approval before proceeding.\n\n⏸ Wait for user.\n\n**After user confirms, add an execution cell to the notebook. Do NOT run the transformation directly (no bash, no inline python). If notebook execution tools (`run_cell`) are available, offer to run the cells. Otherwise, generate the cell for the user to execute themselves:**\n\nIf local execution:\n\n- Add a cell that runs the transformation by importing from the `.py` files already on disk (written by the agent during Steps 7 and 9): import `transform_dataset` from `transform_fn`, load the dataset, transform, and save output. Scripts are located in `\u003Cproject-dir>\u002Fscripts\u002F`.\n\nIf SageMaker Processing Job:\n\n- Add a cell that submits and monitors the Processing Job inline using the V3 SageMaker SDK directly (FrameworkProcessor, ProcessingInput, ProcessingOutput, etc.). Create a FrameworkProcessor with the SKLearn 1.2-1 image, configure inputs\u002Foutputs, and call `processor.run(wait=True, logs=True)` to block the cell and stream logs until the job completes. See `scripts\u002Ftransformation_tools.py` for reference implementation details.\n- Inform the user they can run this cell to kick off and monitor the job.\n\n**Important:** The agent must NOT execute the transformation directly via bash or inline python. If `run_cell` is available, use it to run the notebook cells. Otherwise, the cells are for the user to review and run. Only sample data (from Steps 7 and 9) should be transformed by the agent for validation purposes.\n\n> If `run_cell` is available: \"I've added the execution cell to the notebook. Would you like me to run it?\"\n> Otherwise: \"I've added the execution cell to the notebook. You can run it to transform the full dataset. Would you like to review the notebook before running it?\"\n\n⏸ Wait for user.\n\n### Step 11: Verify and confirm with the user\n\nFor this step, you need: **to verify the output looks correct and confirm with the user.**\n\n- Read 1–2 sample records from the output to show the user.\n- Report the total number of records transformed.\n- Ask the user if the output looks good.\n\n⏸ Wait for user to confirm.\n",{"data":44,"body":47},{"name":4,"description":6,"metadata":45},{"version":46},"1.0.0",{"type":48,"children":49},"root",[50,59,65,72,92,98,115,121,214,220,239,246,272,278,290,326,331,337,343,360,369,392,397,402,408,413,421,425,431,443,451,456,479,487,492,496,502,514,522,526,532,537,542,546,552,563,571,592,596,602,611,624,629,649,669,682,687,697,734,744,748,754,766,770,776,784,809,822,827,868,879,888,923,932,936,942,947,975,983,996,1001,1006,1014,1019,1027,1032,1036,1052,1057,1093,1097,1126,1143,1158,1162,1168,1177,1195,1200],{"type":51,"tag":52,"props":53,"children":55},"element","h1",{"id":54},"dataset-transformation-agent",[56],{"type":57,"value":58},"text","Dataset Transformation Agent",{"type":51,"tag":60,"props":61,"children":62},"p",{},[63],{"type":57,"value":64},"Transforms a data set provided by the user into their desired format.",{"type":51,"tag":66,"props":67,"children":69},"h2",{"id":68},"when-to-use",[70],{"type":57,"value":71},"When to Use",{"type":51,"tag":73,"props":74,"children":75},"ul",{},[76,82,87],{"type":51,"tag":77,"props":78,"children":79},"li",{},[80],{"type":57,"value":81},"User needs to generate code for transforming datasets for SageMaker model training or model evaluation.",{"type":51,"tag":77,"props":83,"children":84},{},[85],{"type":57,"value":86},"A dataset requires processing, cleaning, or formatting before training or evaluation.",{"type":51,"tag":77,"props":88,"children":89},{},[90],{"type":57,"value":91},"Workflow requires a formal review and approval cycle before execution.",{"type":51,"tag":66,"props":93,"children":95},{"id":94},"prerequisites",[96],{"type":57,"value":97},"Prerequisites",{"type":51,"tag":73,"props":99,"children":100},{},[101],{"type":51,"tag":77,"props":102,"children":103},{},[104,106,113],{"type":57,"value":105},"The SDK environment has been verified (SDK version, region, execution role). If not done, activate the ",{"type":51,"tag":107,"props":108,"children":110},"code",{"className":109},[],[111],{"type":57,"value":112},"sdk-getting-started",{"type":57,"value":114}," skill first.",{"type":51,"tag":66,"props":116,"children":118},{"id":117},"principles",[119],{"type":57,"value":120},"Principles",{"type":51,"tag":122,"props":123,"children":124},"ol",{},[125,136,146,156,166,176,186,196],{"type":51,"tag":77,"props":126,"children":127},{},[128,134],{"type":51,"tag":129,"props":130,"children":131},"strong",{},[132],{"type":57,"value":133},"One thing at a time.",{"type":57,"value":135}," Each response advances exactly one decision. Never combine multiple questions or recommendations in a single turn.",{"type":51,"tag":77,"props":137,"children":138},{},[139,144],{"type":51,"tag":129,"props":140,"children":141},{},[142],{"type":57,"value":143},"Confirm before proceeding.",{"type":57,"value":145}," Wait for the user to agree before moving to the next step. You are a guide, not a runaway train.",{"type":51,"tag":77,"props":147,"children":148},{},[149,154],{"type":51,"tag":129,"props":150,"children":151},{},[152],{"type":57,"value":153},"Don't read files until you need them.",{"type":57,"value":155}," Only read reference files when you've reached the workflow step that requires them and the user has confirmed the direction. Never read ahead.",{"type":51,"tag":77,"props":157,"children":158},{},[159,164],{"type":51,"tag":129,"props":160,"children":161},{},[162],{"type":57,"value":163},"No narration.",{"type":57,"value":165}," Don't explain what you're about to do or what you just did. Share outcomes and ask questions. Keep responses short and focused.",{"type":51,"tag":77,"props":167,"children":168},{},[169,174],{"type":51,"tag":129,"props":170,"children":171},{},[172],{"type":57,"value":173},"No repetition.",{"type":57,"value":175}," If you said something before a tool call, don't repeat it after. Only share new information.",{"type":51,"tag":77,"props":177,"children":178},{},[179,184],{"type":51,"tag":129,"props":180,"children":181},{},[182],{"type":57,"value":183},"Do not deviate from the Workflow.",{"type":57,"value":185}," The steps listed in the workflow should be followed exactly as described. Progress from Step 1 to Step 11 to complete the task. Do not deviate from the workflow!",{"type":51,"tag":77,"props":187,"children":188},{},[189,194],{"type":51,"tag":129,"props":190,"children":191},{},[192],{"type":57,"value":193},"Always end with a question.",{"type":57,"value":195}," Whenever you pause for user input, acknowledgment, or feedback, your response must end with a question. Never leave the user with a statement and expect them to know they need to respond.",{"type":51,"tag":77,"props":197,"children":198},{},[199,204,206,212],{"type":51,"tag":129,"props":200,"children":201},{},[202],{"type":57,"value":203},"Default output format is JSONL.",{"type":57,"value":205}," Unless the user explicitly requests a different file format, the transformed dataset should be written as ",{"type":51,"tag":107,"props":207,"children":209},{"className":208},[],[210],{"type":57,"value":211},".jsonl",{"type":57,"value":213}," (JSON Lines — one JSON object per line).",{"type":51,"tag":66,"props":215,"children":217},{"id":216},"known-dataset-formats-reference",[218],{"type":57,"value":219},"Known Dataset Formats Reference",{"type":51,"tag":60,"props":221,"children":222},{},[223,225,230,232,237],{"type":57,"value":224},"This skill supports two transformation purposes — ",{"type":51,"tag":129,"props":226,"children":227},{},[228],{"type":57,"value":229},"training data",{"type":57,"value":231}," and ",{"type":51,"tag":129,"props":233,"children":234},{},[235],{"type":57,"value":236},"evaluation data",{"type":57,"value":238}," — each with its own format resolution path. The purpose is determined in Step 1 of the workflow.",{"type":51,"tag":240,"props":241,"children":243},"h3",{"id":242},"training-data-formats",[244],{"type":57,"value":245},"Training Data Formats",{"type":51,"tag":60,"props":247,"children":248},{},[249,251,256,258,263,265,270],{"type":57,"value":250},"Resolve the target format using the reference file ..\u002Fdataset-evaluation\u002Freferences\u002Fstrategy_data_requirements.md. When the transformation is for ",{"type":51,"tag":129,"props":252,"children":253},{},[254],{"type":57,"value":255},"model training",{"type":57,"value":257},", the required format depends on both the ",{"type":51,"tag":129,"props":259,"children":260},{},[261],{"type":57,"value":262},"model type",{"type":57,"value":264}," (Open Weights like Llama\u002FQwen vs Nova) and the ",{"type":51,"tag":129,"props":266,"children":267},{},[268],{"type":57,"value":269},"finetuning technique",{"type":57,"value":271}," (SFT, DPO, RLVR, RLAIF) — make sure to match on both dimensions. If either the model type or technique is not yet known, ask the user before resolving the format.",{"type":51,"tag":240,"props":273,"children":275},{"id":274},"evaluation-data-formats",[276],{"type":57,"value":277},"Evaluation Data Formats",{"type":51,"tag":60,"props":279,"children":280},{},[281,283,288],{"type":57,"value":282},"When the transformation is for ",{"type":51,"tag":129,"props":284,"children":285},{},[286],{"type":57,"value":287},"model evaluation",{"type":57,"value":289},", resolve the target format using this order:",{"type":51,"tag":122,"props":291,"children":292},{},[293,308],{"type":51,"tag":77,"props":294,"children":295},{},[296,298,306],{"type":57,"value":297},"Try fetching the live documentation at ",{"type":51,"tag":299,"props":300,"children":304},"a",{"href":301,"rel":302},"https:\u002F\u002Fdocs.aws.amazon.com\u002Fsagemaker\u002Flatest\u002Fdg\u002Fmodel-customize-evaluation-dataset-formats.html",[303],"nofollow",[305],{"type":57,"value":301},{"type":57,"value":307}," to get the latest evaluation dataset schema definitions.",{"type":51,"tag":77,"props":309,"children":310},{},[311,316,318,324],{"type":51,"tag":129,"props":312,"children":313},{},[314],{"type":57,"value":315},"If the fetch fails",{"type":57,"value":317}," (e.g., no internet access, VPC environment), fall back to the offline copy at ",{"type":51,"tag":107,"props":319,"children":321},{"className":320},[],[322],{"type":57,"value":323},"references\u002Fsagemaker_dataset_formats.md",{"type":57,"value":325},". Inform the user that the format schemas are from an offline copy and may be outdated.",{"type":51,"tag":60,"props":327,"children":328},{},[329],{"type":57,"value":330},"Use whichever source you successfully access as the source of truth for the target format. Do not rely on memorized schemas.",{"type":51,"tag":66,"props":332,"children":334},{"id":333},"workflow",[335],{"type":57,"value":336},"Workflow",{"type":51,"tag":240,"props":338,"children":340},{"id":339},"step-1-determine-transformation-purpose",[341],{"type":57,"value":342},"Step 1: Determine transformation purpose",{"type":51,"tag":60,"props":344,"children":345},{},[346,348,352,354,358],{"type":57,"value":347},"Your first response should determine whether this transformation is for ",{"type":51,"tag":129,"props":349,"children":350},{},[351],{"type":57,"value":255},{"type":57,"value":353}," or ",{"type":51,"tag":129,"props":355,"children":356},{},[357],{"type":57,"value":287},{"type":57,"value":359},". If the context already makes this clear (e.g., the user said \"I need to prep my training data\" or \"I need to format my eval dataset\"), confirm your understanding and move on. Otherwise, ask:",{"type":51,"tag":361,"props":362,"children":363},"blockquote",{},[364],{"type":51,"tag":60,"props":365,"children":366},{},[367],{"type":57,"value":368},"\"Is this dataset transformation for model training or model evaluation? This helps me look up the right target format for you.\"",{"type":51,"tag":73,"props":370,"children":371},{},[372,382],{"type":51,"tag":77,"props":373,"children":374},{},[375,380],{"type":51,"tag":129,"props":376,"children":377},{},[378],{"type":57,"value":379},"Training",{"type":57,"value":381}," → format resolution will use the local training data requirements reference (model type + finetuning technique dependent).",{"type":51,"tag":77,"props":383,"children":384},{},[385,390],{"type":51,"tag":129,"props":386,"children":387},{},[388],{"type":57,"value":389},"Evaluation",{"type":57,"value":391}," → format resolution will use the live AWS documentation (with offline fallback).",{"type":51,"tag":60,"props":393,"children":394},{},[395],{"type":57,"value":396},"Remember this choice — it determines how the target format is resolved in Step 3.",{"type":51,"tag":60,"props":398,"children":399},{},[400],{"type":57,"value":401},"⏸ Wait for user.",{"type":51,"tag":240,"props":403,"children":405},{"id":404},"step-2-set-expectations",[406],{"type":57,"value":407},"Step 2: Set expectations",{"type":51,"tag":60,"props":409,"children":410},{},[411],{"type":57,"value":412},"Acknowledge the user's request and state what this skill can do:",{"type":51,"tag":361,"props":414,"children":415},{},[416],{"type":51,"tag":60,"props":417,"children":418},{},[419],{"type":57,"value":420},"\"I can help you transform your dataset's format! Here's my plan: I will first need to understand the format of your dataset and the transformation requirements. Once I have that, I will generate a dataset transformation function that we can refine together. After the dataset transformation function is refined to your liking, I will perform the transformation task and upload it to your desired location! Does this sound good?\"",{"type":51,"tag":60,"props":422,"children":423},{},[424],{"type":57,"value":401},{"type":51,"tag":240,"props":426,"children":428},{"id":427},"step-3-understand-the-dataset-transformation-task",[429],{"type":57,"value":430},"Step 3: Understand the dataset transformation task",{"type":51,"tag":60,"props":432,"children":433},{},[434,436,441],{"type":57,"value":435},"For this step, you need to know: ",{"type":51,"tag":129,"props":437,"children":438},{},[439],{"type":57,"value":440},"what dataset format the user would like to transform their dataset from and what dataset format they would like to transform it in to.",{"type":57,"value":442},"\nIf you know this already, skip this step. If not, ask the user:",{"type":51,"tag":361,"props":444,"children":445},{},[446],{"type":51,"tag":60,"props":447,"children":448},{},[449],{"type":57,"value":450},"\"What's the dataset format you would like to transform it into?\"",{"type":51,"tag":60,"props":452,"children":453},{},[454],{"type":57,"value":455},"Resolve the target format based on the purpose determined in Step 1:",{"type":51,"tag":73,"props":457,"children":458},{},[459,469],{"type":51,"tag":77,"props":460,"children":461},{},[462,467],{"type":51,"tag":129,"props":463,"children":464},{},[465],{"type":57,"value":466},"If training data",{"type":57,"value":468},": Ask the user for the finetuning technique (SFT, DPO, RLVR, RLAIF) and model type (Open Weights like Llama\u002FQwen vs Nova) if not already known. Then look up the required format from the \"Training Data Formats\" section in the Known Dataset Formats Reference above.",{"type":51,"tag":77,"props":470,"children":471},{},[472,477],{"type":51,"tag":129,"props":473,"children":474},{},[475],{"type":57,"value":476},"If evaluation data",{"type":57,"value":478},": If the user mentions a well-known format name (e.g., \"OpenAI format\", \"SageMaker format\"), fetch the schema from the live documentation as described in the \"Evaluation Data Formats\" section above. If a well-known format is fetched, confirm with the user:",{"type":51,"tag":361,"props":480,"children":481},{},[482],{"type":51,"tag":60,"props":483,"children":484},{},[485],{"type":57,"value":486},"\"I've found a SageMaker dataset format: {sagemaker-dataset-format-name} with schema: {sagemaker-dataset-format-schema}. Is this what you were referring to?\"",{"type":51,"tag":60,"props":488,"children":489},{},[490],{"type":57,"value":491},"If the user describes a custom format not listed in the reference doc, ask them to provide a sample record of the desired output format.",{"type":51,"tag":60,"props":493,"children":494},{},[495],{"type":57,"value":401},{"type":51,"tag":240,"props":497,"children":499},{"id":498},"step-4-get-the-dataset-from-the-user",[500],{"type":57,"value":501},"Step 4: Get the dataset from the user",{"type":51,"tag":60,"props":503,"children":504},{},[505,507,512],{"type":57,"value":506},"For this step, you need: ",{"type":51,"tag":129,"props":508,"children":509},{},[510],{"type":57,"value":511},"the location of the user's dataset",{"type":57,"value":513},".\nIf you know this already, skip this step. If not, ask the user:",{"type":51,"tag":361,"props":515,"children":516},{},[517],{"type":51,"tag":60,"props":518,"children":519},{},[520],{"type":57,"value":521},"\"Where can I find your dataset? Either a local directory or S3 location works!\"",{"type":51,"tag":60,"props":523,"children":524},{},[525],{"type":57,"value":401},{"type":51,"tag":240,"props":527,"children":529},{"id":528},"step-5-examine-sample-data",[530],{"type":57,"value":531},"Step 5: Examine sample data",{"type":51,"tag":60,"props":533,"children":534},{},[535],{"type":57,"value":536},"Read 1–2 sample records from the user's dataset and show them so the user can confirm the source schema. Do not run format detection — that is handled by the planning skill before this skill is invoked.",{"type":51,"tag":60,"props":538,"children":539},{},[540],{"type":57,"value":541},"Do not show a side-by-side mapping to the target format here — the detailed mapping will be handled in Step 7 when generating the transformation function.",{"type":51,"tag":60,"props":543,"children":544},{},[545],{"type":57,"value":401},{"type":51,"tag":240,"props":547,"children":549},{"id":548},"step-6-get-the-dataset-output-location",[550],{"type":57,"value":551},"Step 6: Get the dataset output location",{"type":51,"tag":60,"props":553,"children":554},{},[555,556,561],{"type":57,"value":506},{"type":51,"tag":129,"props":557,"children":558},{},[559],{"type":57,"value":560},"to understand where to output the transformed dataset to. It could be an S3 URI or local directory",{"type":57,"value":562},"\nIf you already know where the dataset is supposed to be output to, skip this step. If not, ask the user:",{"type":51,"tag":361,"props":564,"children":565},{},[566],{"type":51,"tag":60,"props":567,"children":568},{},[569],{"type":57,"value":570},"\"Where should I output your transformed dataset to? Either a local directory or S3 location works!\"",{"type":51,"tag":60,"props":572,"children":573},{},[574,576,582,584,590],{"type":57,"value":575},"If the user provides a directory (not a full file path), construct the output filename using the pattern ",{"type":51,"tag":107,"props":577,"children":579},{"className":578},[],[580],{"type":57,"value":581},"{original_name}_{target_format}.jsonl",{"type":57,"value":583}," (e.g., ",{"type":51,"tag":107,"props":585,"children":587},{"className":586},[],[588],{"type":57,"value":589},"gen_qa_100k_openai.jsonl",{"type":57,"value":591},").",{"type":51,"tag":60,"props":593,"children":594},{},[595],{"type":57,"value":401},{"type":51,"tag":240,"props":597,"children":599},{"id":598},"step-7-generate-and-validate-the-transformation-function",[600],{"type":57,"value":601},"Step 7: Generate and validate the transformation function",{"type":51,"tag":60,"props":603,"children":604},{},[605,606],{"type":57,"value":506},{"type":51,"tag":129,"props":607,"children":608},{},[609],{"type":57,"value":610},"to generate a python function that transforms the dataset from the format in Step 5 to the format in Step 3",{"type":51,"tag":60,"props":612,"children":613},{},[614,616,622],{"type":57,"value":615},"Read the reference guide at ",{"type":51,"tag":107,"props":617,"children":619},{"className":618},[],[620],{"type":57,"value":621},"references\u002Fdataset_transformation_code.md",{"type":57,"value":623}," and follow its skeleton exactly when generating the transformation function.",{"type":51,"tag":60,"props":625,"children":626},{},[627],{"type":57,"value":628},"The python function should be in the form of:",{"type":51,"tag":630,"props":631,"children":635},"pre",{"className":632,"code":633,"language":25,"meta":634,"style":634},"language-python shiki shiki-themes material-theme-lighter material-theme material-theme-palenight","def transform_dataset(df: pd.DataFrame) -> pd.DataFrame:\n","",[636],{"type":51,"tag":107,"props":637,"children":638},{"__ignoreMap":634},[639],{"type":51,"tag":640,"props":641,"children":644},"span",{"class":642,"line":643},"line",1,[645],{"type":51,"tag":640,"props":646,"children":647},{},[648],{"type":57,"value":633},{"type":51,"tag":60,"props":650,"children":651},{},[652,654,660,662,668],{"type":57,"value":653},"The ",{"type":51,"tag":107,"props":655,"children":657},{"className":656},[],[658],{"type":57,"value":659},"\u003Cproject-dir>",{"type":57,"value":661}," is the project directory established by the directory-management skill (e.g., ",{"type":51,"tag":107,"props":663,"children":665},{"className":664},[],[666],{"type":57,"value":667},"dpo-to-rlvr-conversion",{"type":57,"value":591},{"type":51,"tag":60,"props":670,"children":671},{},[672,674,680],{"type":57,"value":673},"In notebook mode, add a ",{"type":51,"tag":107,"props":675,"children":677},{"className":676},[],[678],{"type":57,"value":679},"%%writefile \u003Cproject-dir>\u002Fscripts\u002Ftransform_fn.py",{"type":57,"value":681}," code cell AND write the file to disk for testing. In script mode, write the file to disk directly.",{"type":51,"tag":60,"props":683,"children":684},{},[685],{"type":57,"value":686},"Continue iterating with the user's feedback — update the code in place on each revision rather than showing code inline.",{"type":51,"tag":60,"props":688,"children":689},{},[690,695],{"type":51,"tag":129,"props":691,"children":692},{},[693],{"type":57,"value":694},"If sample data was collected in Step 5",{"type":57,"value":696},", test the function against the sample records:",{"type":51,"tag":122,"props":698,"children":699},{},[700,705,724,729],{"type":51,"tag":77,"props":701,"children":702},{},[703],{"type":57,"value":704},"Generate the transformation function.",{"type":51,"tag":77,"props":706,"children":707},{},[708,710,716,718],{"type":57,"value":709},"Write the sample data to a temporary JSONL file (e.g., ",{"type":51,"tag":107,"props":711,"children":713},{"className":712},[],[714],{"type":57,"value":715},"\u002Ftmp\u002Ftest_input.jsonl",{"type":57,"value":717},"), then run:\n",{"type":51,"tag":107,"props":719,"children":721},{"className":720},[],[722],{"type":57,"value":723},"python3 -c \"import sys; sys.path.insert(0, '\u003Cproject-dir>\u002Fscripts'); from transform_fn import transform_dataset; import pandas as pd; df = pd.read_json('\u002Ftmp\u002Ftest_input.jsonl', lines=True); result = transform_dataset(df); print(result.to_json(orient='records', lines=True))\"",{"type":51,"tag":77,"props":725,"children":726},{},[727],{"type":57,"value":728},"If the test fails, fix and re-test until it passes.",{"type":51,"tag":77,"props":730,"children":731},{},[732],{"type":57,"value":733},"Show the user the function and transformed sample output for review.",{"type":51,"tag":60,"props":735,"children":736},{},[737,742],{"type":51,"tag":129,"props":738,"children":739},{},[740],{"type":57,"value":741},"If no sample data",{"type":57,"value":743},", present the function for review and refinement.",{"type":51,"tag":60,"props":745,"children":746},{},[747],{"type":57,"value":401},{"type":51,"tag":240,"props":749,"children":751},{"id":750},"step-8-determine-output-target",[752],{"type":57,"value":753},"Step 8: Determine output target",{"type":51,"tag":60,"props":755,"children":756},{},[757,759,764],{"type":57,"value":758},"If no project directory exists, activate the ",{"type":51,"tag":129,"props":760,"children":761},{},[762],{"type":57,"value":763},"directory-management",{"type":57,"value":765}," skill to set one up.",{"type":51,"tag":60,"props":767,"children":768},{},[769],{"type":57,"value":401},{"type":51,"tag":240,"props":771,"children":773},{"id":772},"step-9-generate-the-execution-code",[774],{"type":57,"value":775},"Step 9: Generate the execution code",{"type":51,"tag":60,"props":777,"children":778},{},[779],{"type":51,"tag":129,"props":780,"children":781},{},[782],{"type":57,"value":783},"Before writing the code, read:",{"type":51,"tag":73,"props":785,"children":786},{},[787,798],{"type":51,"tag":77,"props":788,"children":789},{},[790,796],{"type":51,"tag":107,"props":791,"children":793},{"className":792},[],[794],{"type":57,"value":795},"references\u002Fcode_output_guide.md",{"type":57,"value":797}," (output format rules)",{"type":51,"tag":77,"props":799,"children":800},{},[801,807],{"type":51,"tag":107,"props":802,"children":804},{"className":803},[],[805],{"type":57,"value":806},"code_templates\u002Ftransformation.py",{"type":57,"value":808}," (cell structure and skeleton code)",{"type":51,"tag":60,"props":810,"children":811},{},[812,814,820],{"type":57,"value":813},"The template uses ",{"type":51,"tag":107,"props":815,"children":817},{"className":816},[],[818],{"type":57,"value":819},"# Cell N: Label",{"type":57,"value":821}," markers — each marker starts a new section. Cell 2 (Transformation Function) is dynamically generated from Step 7; all other cells follow the template skeleton.",{"type":51,"tag":60,"props":823,"children":824},{},[825],{"type":57,"value":826},"Generate the execution logic following the code output guide.",{"type":51,"tag":73,"props":828,"children":829},{},[830,842,863],{"type":51,"tag":77,"props":831,"children":832},{},[833,834,840],{"type":57,"value":673},{"type":51,"tag":107,"props":835,"children":837},{"className":836},[],[838],{"type":57,"value":839},"%%writefile \u003Cproject-dir>\u002Fscripts\u002F\u003Cscript_name>.py",{"type":57,"value":841}," code cell AND write the file to disk. In script mode, write the file to disk directly.",{"type":51,"tag":77,"props":843,"children":844},{},[845,847,853,855,861],{"type":57,"value":846},"The script must import ",{"type":51,"tag":107,"props":848,"children":850},{"className":849},[],[851],{"type":57,"value":852},"transform_dataset",{"type":57,"value":854}," from ",{"type":51,"tag":107,"props":856,"children":858},{"className":857},[],[859],{"type":57,"value":860},"transform_fn",{"type":57,"value":862},".",{"type":51,"tag":77,"props":864,"children":865},{},[866],{"type":57,"value":867},"Replace placeholders with the actual input\u002Foutput paths.",{"type":51,"tag":60,"props":869,"children":870},{},[871,872,877],{"type":57,"value":615},{"type":51,"tag":107,"props":873,"children":875},{"className":874},[],[876],{"type":57,"value":621},{"type":57,"value":878}," and follow its execution script skeleton exactly.",{"type":51,"tag":60,"props":880,"children":881},{},[882,886],{"type":51,"tag":129,"props":883,"children":884},{},[885],{"type":57,"value":694},{"type":57,"value":887},", test the full pipeline:",{"type":51,"tag":122,"props":889,"children":890},{},[891,902,913,918],{"type":51,"tag":77,"props":892,"children":893},{},[894,896,901],{"type":57,"value":895},"Write the sample records to a temporary JSONL file (e.g., ",{"type":51,"tag":107,"props":897,"children":899},{"className":898},[],[900],{"type":57,"value":715},{"type":57,"value":591},{"type":51,"tag":77,"props":903,"children":904},{},[905,907],{"type":57,"value":906},"Run: ",{"type":51,"tag":107,"props":908,"children":910},{"className":909},[],[911],{"type":57,"value":912},"python3 \u003Cproject-dir>\u002Fscripts\u002F\u003Cscript_name> --input \u002Ftmp\u002Ftest_input.jsonl --output \u002Ftmp\u002Ftest_output.jsonl",{"type":51,"tag":77,"props":914,"children":915},{},[916],{"type":57,"value":917},"If it fails, debug and fix, then re-run until successful.",{"type":51,"tag":77,"props":919,"children":920},{},[921],{"type":57,"value":922},"Show the user the output for review.",{"type":51,"tag":60,"props":924,"children":925},{},[926,930],{"type":51,"tag":129,"props":927,"children":928},{},[929],{"type":57,"value":741},{"type":57,"value":931},", present the notebook for review and refinement.",{"type":51,"tag":60,"props":933,"children":934},{},[935],{"type":57,"value":401},{"type":51,"tag":240,"props":937,"children":939},{"id":938},"step-10-determine-and-confirm-execution-mode",[940],{"type":57,"value":941},"Step 10: Determine and confirm execution mode",{"type":51,"tag":60,"props":943,"children":944},{},[945],{"type":57,"value":946},"Check the size of the input dataset:",{"type":51,"tag":73,"props":948,"children":949},{},[950,970],{"type":51,"tag":77,"props":951,"children":952},{},[953,955,961,963,969],{"type":57,"value":954},"If the dataset is in S3, use the AWS MCP tool ",{"type":51,"tag":107,"props":956,"children":958},{"className":957},[],[959],{"type":57,"value":960},"head-object",{"type":57,"value":962}," (S3 service) with the bucket and key to get ",{"type":51,"tag":107,"props":964,"children":966},{"className":965},[],[967],{"type":57,"value":968},"ContentLength",{"type":57,"value":862},{"type":51,"tag":77,"props":971,"children":972},{},[973],{"type":57,"value":974},"If the dataset is local, check the file size.",{"type":51,"tag":60,"props":976,"children":977},{},[978],{"type":51,"tag":129,"props":979,"children":980},{},[981],{"type":57,"value":982},"Decision criteria:",{"type":51,"tag":73,"props":984,"children":985},{},[986,991],{"type":51,"tag":77,"props":987,"children":988},{},[989],{"type":57,"value":990},"Dataset \u003C 50 MB → recommend local execution",{"type":51,"tag":77,"props":992,"children":993},{},[994],{"type":57,"value":995},"Dataset ≥ 50 MB → recommend SageMaker Processing Job",{"type":51,"tag":60,"props":997,"children":998},{},[999],{"type":57,"value":1000},"Inform the user of the recommendation and get their approval:",{"type":51,"tag":60,"props":1002,"children":1003},{},[1004],{"type":57,"value":1005},"If local:",{"type":51,"tag":361,"props":1007,"children":1008},{},[1009],{"type":51,"tag":60,"props":1010,"children":1011},{},[1012],{"type":57,"value":1013},"\"Your dataset is {size} MB — since it's under 50 MB, I'd recommend running the transformation locally. Would you like to proceed with local execution, or would you prefer a SageMaker Processing Job instead?\"",{"type":51,"tag":60,"props":1015,"children":1016},{},[1017],{"type":57,"value":1018},"If SageMaker Processing Job:",{"type":51,"tag":361,"props":1020,"children":1021},{},[1022],{"type":51,"tag":60,"props":1023,"children":1024},{},[1025],{"type":57,"value":1026},"\"Your dataset is {size} MB — since it's over 50 MB, I'd recommend running this as a SageMaker Processing Job for better performance. Would you like to proceed with a SageMaker Processing Job, or would you prefer to run it locally instead?\"",{"type":51,"tag":60,"props":1028,"children":1029},{},[1030],{"type":57,"value":1031},"Do not execute until the user approves. If the user rejects the recommendation, switch to the alternative and get their explicit approval before proceeding.",{"type":51,"tag":60,"props":1033,"children":1034},{},[1035],{"type":57,"value":401},{"type":51,"tag":60,"props":1037,"children":1038},{},[1039],{"type":51,"tag":129,"props":1040,"children":1041},{},[1042,1044,1050],{"type":57,"value":1043},"After user confirms, add an execution cell to the notebook. Do NOT run the transformation directly (no bash, no inline python). If notebook execution tools (",{"type":51,"tag":107,"props":1045,"children":1047},{"className":1046},[],[1048],{"type":57,"value":1049},"run_cell",{"type":57,"value":1051},") are available, offer to run the cells. Otherwise, generate the cell for the user to execute themselves:",{"type":51,"tag":60,"props":1053,"children":1054},{},[1055],{"type":57,"value":1056},"If local execution:",{"type":51,"tag":73,"props":1058,"children":1059},{},[1060],{"type":51,"tag":77,"props":1061,"children":1062},{},[1063,1065,1071,1073,1078,1079,1084,1086,1092],{"type":57,"value":1064},"Add a cell that runs the transformation by importing from the ",{"type":51,"tag":107,"props":1066,"children":1068},{"className":1067},[],[1069],{"type":57,"value":1070},".py",{"type":57,"value":1072}," files already on disk (written by the agent during Steps 7 and 9): import ",{"type":51,"tag":107,"props":1074,"children":1076},{"className":1075},[],[1077],{"type":57,"value":852},{"type":57,"value":854},{"type":51,"tag":107,"props":1080,"children":1082},{"className":1081},[],[1083],{"type":57,"value":860},{"type":57,"value":1085},", load the dataset, transform, and save output. Scripts are located in ",{"type":51,"tag":107,"props":1087,"children":1089},{"className":1088},[],[1090],{"type":57,"value":1091},"\u003Cproject-dir>\u002Fscripts\u002F",{"type":57,"value":862},{"type":51,"tag":60,"props":1094,"children":1095},{},[1096],{"type":57,"value":1018},{"type":51,"tag":73,"props":1098,"children":1099},{},[1100,1121],{"type":51,"tag":77,"props":1101,"children":1102},{},[1103,1105,1111,1113,1119],{"type":57,"value":1104},"Add a cell that submits and monitors the Processing Job inline using the V3 SageMaker SDK directly (FrameworkProcessor, ProcessingInput, ProcessingOutput, etc.). Create a FrameworkProcessor with the SKLearn 1.2-1 image, configure inputs\u002Foutputs, and call ",{"type":51,"tag":107,"props":1106,"children":1108},{"className":1107},[],[1109],{"type":57,"value":1110},"processor.run(wait=True, logs=True)",{"type":57,"value":1112}," to block the cell and stream logs until the job completes. See ",{"type":51,"tag":107,"props":1114,"children":1116},{"className":1115},[],[1117],{"type":57,"value":1118},"scripts\u002Ftransformation_tools.py",{"type":57,"value":1120}," for reference implementation details.",{"type":51,"tag":77,"props":1122,"children":1123},{},[1124],{"type":57,"value":1125},"Inform the user they can run this cell to kick off and monitor the job.",{"type":51,"tag":60,"props":1127,"children":1128},{},[1129,1134,1136,1141],{"type":51,"tag":129,"props":1130,"children":1131},{},[1132],{"type":57,"value":1133},"Important:",{"type":57,"value":1135}," The agent must NOT execute the transformation directly via bash or inline python. If ",{"type":51,"tag":107,"props":1137,"children":1139},{"className":1138},[],[1140],{"type":57,"value":1049},{"type":57,"value":1142}," is available, use it to run the notebook cells. Otherwise, the cells are for the user to review and run. Only sample data (from Steps 7 and 9) should be transformed by the agent for validation purposes.",{"type":51,"tag":361,"props":1144,"children":1145},{},[1146],{"type":51,"tag":60,"props":1147,"children":1148},{},[1149,1151,1156],{"type":57,"value":1150},"If ",{"type":51,"tag":107,"props":1152,"children":1154},{"className":1153},[],[1155],{"type":57,"value":1049},{"type":57,"value":1157}," is available: \"I've added the execution cell to the notebook. Would you like me to run it?\"\nOtherwise: \"I've added the execution cell to the notebook. You can run it to transform the full dataset. Would you like to review the notebook before running it?\"",{"type":51,"tag":60,"props":1159,"children":1160},{},[1161],{"type":57,"value":401},{"type":51,"tag":240,"props":1163,"children":1165},{"id":1164},"step-11-verify-and-confirm-with-the-user",[1166],{"type":57,"value":1167},"Step 11: Verify and confirm with the user",{"type":51,"tag":60,"props":1169,"children":1170},{},[1171,1172],{"type":57,"value":506},{"type":51,"tag":129,"props":1173,"children":1174},{},[1175],{"type":57,"value":1176},"to verify the output looks correct and confirm with the user.",{"type":51,"tag":73,"props":1178,"children":1179},{},[1180,1185,1190],{"type":51,"tag":77,"props":1181,"children":1182},{},[1183],{"type":57,"value":1184},"Read 1–2 sample records from the output to show the user.",{"type":51,"tag":77,"props":1186,"children":1187},{},[1188],{"type":57,"value":1189},"Report the total number of records transformed.",{"type":51,"tag":77,"props":1191,"children":1192},{},[1193],{"type":57,"value":1194},"Ask the user if the output looks good.",{"type":51,"tag":60,"props":1196,"children":1197},{},[1198],{"type":57,"value":1199},"⏸ Wait for user to confirm.",{"type":51,"tag":1201,"props":1202,"children":1203},"style",{},[1204],{"type":57,"value":1205},"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":1207,"total":1386},[1208,1228,1249,1259,1272,1285,1295,1305,1326,1341,1356,1371],{"slug":1209,"name":1209,"fn":1210,"description":1211,"org":1212,"tags":1213,"stars":1225,"repoUrl":1226,"updatedAt":1227},"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},[1214,1216,1219,1222],{"name":1215,"slug":35,"type":16},"AWS",{"name":1217,"slug":1218,"type":16},"Debugging","debugging",{"name":1220,"slug":1221,"type":16},"Logs","logs",{"name":1223,"slug":1224,"type":16},"Observability","observability",9427,"https:\u002F\u002Fgithub.com\u002Fawslabs\u002Fmcp","2026-07-12T08:37:22.601527",{"slug":1229,"name":1230,"fn":1231,"description":1232,"org":1233,"tags":1234,"stars":1225,"repoUrl":1226,"updatedAt":1248},"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},[1235,1238,1239,1242,1245],{"name":1236,"slug":1237,"type":16},"Aurora","aurora",{"name":1215,"slug":35,"type":16},{"name":1240,"slug":1241,"type":16},"Database","database",{"name":1243,"slug":1244,"type":16},"Serverless","serverless",{"name":1246,"slug":1247,"type":16},"SQL","sql","2026-07-12T08:36:45.053393",{"slug":1250,"name":1251,"fn":1231,"description":1232,"org":1252,"tags":1253,"stars":1225,"repoUrl":1226,"updatedAt":1258},"aurora-dsql","aurora dsql",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1254,1255,1256,1257],{"name":1215,"slug":35,"type":16},{"name":1240,"slug":1241,"type":16},{"name":1243,"slug":1244,"type":16},{"name":1246,"slug":1247,"type":16},"2026-07-12T08:36:42.694299",{"slug":1260,"name":1261,"fn":1231,"description":1232,"org":1262,"tags":1263,"stars":1225,"repoUrl":1226,"updatedAt":1271},"aws-dsql","aws dsql",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1264,1265,1266,1269,1270],{"name":1215,"slug":35,"type":16},{"name":1240,"slug":1241,"type":16},{"name":1267,"slug":1268,"type":16},"Migration","migration",{"name":1243,"slug":1244,"type":16},{"name":1246,"slug":1247,"type":16},"2026-07-12T08:36:38.584057",{"slug":1273,"name":1274,"fn":1231,"description":1232,"org":1275,"tags":1276,"stars":1225,"repoUrl":1226,"updatedAt":1284},"distributed-postgres","distributed postgres",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1277,1278,1279,1282,1283],{"name":1215,"slug":35,"type":16},{"name":1240,"slug":1241,"type":16},{"name":1280,"slug":1281,"type":16},"PostgreSQL","postgresql",{"name":1243,"slug":1244,"type":16},{"name":1246,"slug":1247,"type":16},"2026-07-12T08:36:46.530743",{"slug":1286,"name":1287,"fn":1231,"description":1232,"org":1288,"tags":1289,"stars":1225,"repoUrl":1226,"updatedAt":1294},"distributed-sql","distributed sql",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1290,1291,1292,1293],{"name":1215,"slug":35,"type":16},{"name":1240,"slug":1241,"type":16},{"name":1243,"slug":1244,"type":16},{"name":1246,"slug":1247,"type":16},"2026-07-12T08:36:48.104182",{"slug":1296,"name":1296,"fn":1231,"description":1232,"org":1297,"tags":1298,"stars":1225,"repoUrl":1226,"updatedAt":1304},"dsql",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[1299,1300,1301,1302,1303],{"name":1215,"slug":35,"type":16},{"name":1240,"slug":1241,"type":16},{"name":1267,"slug":1268,"type":16},{"name":1243,"slug":1244,"type":16},{"name":1246,"slug":1247,"type":16},"2026-07-12T08:36:36.374512",{"slug":1306,"name":1306,"fn":1307,"description":1308,"org":1309,"tags":1310,"stars":1323,"repoUrl":1324,"updatedAt":1325},"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},[1311,1314,1317,1320],{"name":1312,"slug":1313,"type":16},"Accounting","accounting",{"name":1315,"slug":1316,"type":16},"Analytics","analytics",{"name":1318,"slug":1319,"type":16},"Cost Optimization","cost-optimization",{"name":1321,"slug":1322,"type":16},"Finance","finance",3176,"https:\u002F\u002Fgithub.com\u002Fawslabs\u002Fagentcore-samples","2026-07-12T08:40:03.29555",{"slug":1327,"name":1327,"fn":1328,"description":1329,"org":1330,"tags":1331,"stars":1323,"repoUrl":1324,"updatedAt":1340},"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},[1332,1333,1334,1337],{"name":1215,"slug":35,"type":16},{"name":1321,"slug":1322,"type":16},{"name":1335,"slug":1336,"type":16},"Management","management",{"name":1338,"slug":1339,"type":16},"Reporting","reporting","2026-07-12T08:40:02.066471",{"slug":1342,"name":1342,"fn":1343,"description":1344,"org":1345,"tags":1346,"stars":1323,"repoUrl":1324,"updatedAt":1355},"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},[1347,1348,1349,1352],{"name":1315,"slug":1316,"type":16},{"name":1321,"slug":1322,"type":16},{"name":1350,"slug":1351,"type":16},"Financial Statements","financial-statements",{"name":1353,"slug":1354,"type":16},"Variance Analysis","variance-analysis","2026-07-12T08:40:00.79141",{"slug":1357,"name":1357,"fn":1358,"description":1359,"org":1360,"tags":1361,"stars":1323,"repoUrl":1324,"updatedAt":1370},"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},[1362,1365,1368],{"name":1363,"slug":1364,"type":16},"Automation","automation",{"name":1366,"slug":1367,"type":16},"Documents","documents",{"name":1369,"slug":1357,"type":16},"PDF","2026-07-12T08:41:44.135656",{"slug":1372,"name":1372,"fn":1373,"description":1374,"org":1375,"tags":1376,"stars":1323,"repoUrl":1324,"updatedAt":1385},"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},[1377,1378,1381,1382],{"name":1312,"slug":1313,"type":16},{"name":1379,"slug":1380,"type":16},"Data Analysis","data-analysis",{"name":1321,"slug":1322,"type":16},{"name":1383,"slug":1384,"type":16},"KPI","kpi","2026-07-12T08:39:59.54971",150,{"items":1388,"total":1489},[1389,1406,1424,1436,1453,1464,1474],{"slug":1390,"name":1390,"fn":1391,"description":1392,"org":1393,"tags":1394,"stars":26,"repoUrl":27,"updatedAt":1405},"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},[1395,1398,1399,1402],{"name":1396,"slug":1397,"type":16},"API Development","api-development",{"name":1215,"slug":35,"type":16},{"name":1400,"slug":1401,"type":16},"Maps","maps",{"name":1403,"slug":1404,"type":16},"Navigation","navigation","2026-07-12T08:39:49.88311",{"slug":1407,"name":1407,"fn":1408,"description":1409,"org":1410,"tags":1411,"stars":26,"repoUrl":27,"updatedAt":1423},"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},[1412,1415,1416,1417,1420],{"name":1413,"slug":1414,"type":16},"Auth","auth",{"name":1215,"slug":35,"type":16},{"name":1240,"slug":1241,"type":16},{"name":1418,"slug":1419,"type":16},"Deployment","deployment",{"name":1421,"slug":1422,"type":16},"TypeScript","typescript","2026-07-12T08:39:43.500162",{"slug":1425,"name":1425,"fn":1426,"description":1427,"org":1428,"tags":1429,"stars":26,"repoUrl":27,"updatedAt":1435},"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},[1430,1431,1432],{"name":1396,"slug":1397,"type":16},{"name":1215,"slug":35,"type":16},{"name":1433,"slug":1434,"type":16},"REST API","rest-api","2026-07-12T08:39:00.149339",{"slug":1437,"name":1437,"fn":1438,"description":1439,"org":1440,"tags":1441,"stars":26,"repoUrl":27,"updatedAt":1452},"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},[1442,1445,1446,1449],{"name":1443,"slug":1444,"type":16},"Architecture","architecture",{"name":1215,"slug":35,"type":16},{"name":1447,"slug":1448,"type":16},"Design","design",{"name":1450,"slug":1451,"type":16},"Diagrams","diagrams","2026-07-12T08:37:11.012278",{"slug":1454,"name":1454,"fn":1455,"description":1456,"org":1457,"tags":1458,"stars":26,"repoUrl":27,"updatedAt":1463},"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},[1459,1460,1461,1462],{"name":1396,"slug":1397,"type":16},{"name":1215,"slug":35,"type":16},{"name":1418,"slug":1419,"type":16},{"name":1243,"slug":1244,"type":16},"2026-07-12T08:38:58.598492",{"slug":1465,"name":1465,"fn":1466,"description":1467,"org":1468,"tags":1469,"stars":26,"repoUrl":27,"updatedAt":1473},"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},[1470,1471,1472],{"name":1443,"slug":1444,"type":16},{"name":1215,"slug":35,"type":16},{"name":1243,"slug":1244,"type":16},"2026-07-12T08:39:05.546173",{"slug":1475,"name":1475,"fn":1476,"description":1477,"org":1478,"tags":1479,"stars":26,"repoUrl":27,"updatedAt":1488},"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},[1480,1481,1482,1485],{"name":1215,"slug":35,"type":16},{"name":1418,"slug":1419,"type":16},{"name":1483,"slug":1484,"type":16},"Infrastructure","infrastructure",{"name":1486,"slug":1487,"type":16},"Performance","performance","2026-07-12T08:39:07.007071",33]