Open-source agentic data engineering harness.

100+ deterministic tools for SQL, lineage, dbt, FinOps, and warehouse connectivity.

$npm install -g altimate-code
View the project on GitHub

Works with your data stack

SnowflakeSnowflake
BigQueryBigQuery
DatabricksDatabricks
PostgreSQLPostgreSQL
dbtdbt
AirflowAirflow
GitHubGitHub
RedshiftRedshift
MySQLMySQL
JiraJira
LinearLinear
DuckDBDuckDB
DagsterDagster
Google SheetsGoogle Sheets
SnowflakeSnowflake
BigQueryBigQuery
DatabricksDatabricks
PostgreSQLPostgreSQL
dbtdbt
AirflowAirflow
GitHubGitHub
RedshiftRedshift
MySQLMySQL
JiraJira
LinearLinear
DuckDBDuckDB
DagsterDagster
Google SheetsGoogle Sheets

Bring your own LLM

AnthropicAnthropic
OpenAIOpenAI
GoogleGoogle
AWS BedrockAWS Bedrock
AzureAzure
OllamaOllama
OpenRouterOpenRouter
MistralMistral
GroqGroq
xAIxAI
CopilotCopilot
AnthropicAnthropic
OpenAIOpenAI
GoogleGoogle
AWS BedrockAWS Bedrock
AzureAzure
OllamaOllama
OpenRouterOpenRouter
MistralMistral
GroqGroq
xAIxAI
CopilotCopilot

Why altimate-code

Six things a generic coding agent does not give you.

Every capability is open source, MIT-licensed, and reproducible from a published benchmark.

Bring Your Own LLM

Anthropic, OpenAI, Google, Bedrock, Azure, Ollama, Snowflake Cortex, Databricks AI Gateway, and more. No vendor lock-in.

12+ providers · BYO key

Cross-Platform

Claude Code, Cursor, Windsurf, VS Code, and any MCP-compatible client. One install, everywhere.

MCP-native · one install

100+ Deterministic Tools

SQL analysis, column-level lineage, dbt integration, FinOps, warehouse connectivity. No hallucinations.

100+ tools · 34 SQL dialectsRead proof →

Validation Layer

SQL, lineage, and equivalence checks run in compiled Rust, not the model. Same input, same output, every run.

100% F1 · ~2 ms · 0 tokensRead proof →

Token Efficiency

Context compaction trims the schema payload per task. The gateway routes each call to the cheapest model that clears your accuracy bar.

per-task routing · context compactionRead proof →

Data Governance

Policy gates, PII detection, and compliance checks run before the agent writes a column. Audit-ready.

policy · PII · audit trail

See it in action

Real-world pipelines built end-to-end with a single prompt

US Home Sales Data Science Dashboard
$Download all available public US home sales data sets. Process and merge them into a unified format. Perform advanced data science — K-means, OLS regressions, and more. Build a single interactive dashboard.
US Home Sales Data Science Dashboard
Snowflake vs Databricks Benchmark
$Deploy the NovaMart platform to both Snowflake and Databricks, testing multiple warehouse sizes on each. Run the full data pipeline and benchmark queries, capturing execution time, credits/DBUs consumed, and bytes scanned.
Snowflake vs Databricks Benchmark
NYC Taxi Coverage Dashboard
DuckDBdbtAirflow
$Take the New York City taxi cab public dataset, bring up a DuckDB instance, and build a dashboard showing areas of maximum coverage and lowest coverage. Set up a complete dbt project with staging, intermediate, and mart layers, and create an Airflow DAG to orchestrate the pipeline.
NYC Taxi Coverage Dashboard
Olist E-Commerce Analytics Pipeline
$Build an end-to-end e-commerce analytics pipeline using the Olist Brazilian E-Commerce dataset. Use Azure Data Factory to ingest CSV files from Blob Storage into Snowflake raw tables, then orchestrate Snowflake stored procedures to transform data through raw → staging → mart layers.
Olist E-Commerce Analytics Pipeline
Global CO2 & Climate Explorer
$Build me an interactive Global CO2 & Climate Explorer dashboard using DuckDB-WASM running entirely in the browser, sourcing data from Our World in Data’s CO2 dataset.
Global CO2 & Climate Explorer

NYC Taxi Coverage Dashboard

Curious to know more? Check out the examples

Join the Community

altimate-code is built in the open. Get involved.

Open source. MIT licensed. One install.

Frequently Asked Questions