What can dbt-docs do?

dbt project metadata extraction and exposure, model-level lineage graph construction and traversal, column-level lineage and data type tracking, dbt documentation content retrieval and search, test coverage analysis and test metadata exposure, source and external table metadata exposure, mcp protocol integration and tool exposure, manifest.json parsing and caching

dbt-docs

MCP ServerFree

** - MCP server for dbt-core (OSS) users as the official dbt MCP only supports dbt Cloud. Supports project metadata, model and column-level lineage and dbt documentation.

Open Source

/ 100

8 capabilities

Capabilities8 decomposed

dbt project metadata extraction and exposure

Medium confidence

Parses dbt project configuration files (dbt_project.yml, manifest.json) and exposes project-level metadata including model counts, source definitions, test coverage, and documentation status through MCP tools. Implements a manifest-based approach that reads the compiled dbt artifact rather than executing dbt commands, enabling fast metadata queries without project recompilation.

Solves for

Query dbt project structure and model inventory programmaticallyRetrieve project-level statistics like total models, sources, and testsAccess dbt configuration and variable definitions for context in LLM agentsExpose dbt documentation metadata to external tools via MCP protocol

Best for

dbt-core OSS users building LLM agents that need project awareness

Teams integrating dbt metadata into Claude or other MCP-compatible clients

Data engineers automating dbt project analysis and governance

Requires

dbt-core 1.0+ with compiled manifest.json artifact

Python 3.8+ for MCP server runtime

MCP-compatible client (Claude Desktop, Cline, etc.)

Limitations

Requires pre-compiled manifest.json — does not execute dbt commands, so metadata reflects last successful dbt parse/run

No real-time project state — changes to dbt_project.yml require manual dbt parse to update manifest

Limited to dbt-core projects; does not support dbt Cloud-specific features like discovery API

What makes it unique

Operates on pre-compiled dbt artifacts (manifest.json) rather than requiring dbt CLI execution, enabling instant metadata queries without triggering dbt parse/run cycles. Fills the gap for dbt-core users who lack access to the official dbt Cloud MCP.

vs alternatives

Faster and lighter than dbt Cloud MCP for local dbt-core projects because it reads cached artifacts instead of making API calls, and requires no dbt Cloud subscription.

model-level lineage graph construction and traversal

Medium confidence

Reconstructs dbt model dependency graphs from manifest.json by parsing upstream/downstream relationships between models, sources, and tests. Exposes lineage as queryable graph structure enabling traversal of data flow paths, impact analysis, and dependency visualization. Uses manifest node relationships to build directed acyclic graph (DAG) without executing dbt commands.

Solves for

Trace data lineage from source tables through transformations to final modelsIdentify downstream models affected by changes to a specific modelFind upstream sources and dependencies for a given modelAnalyze impact of schema changes or source modifications on dependent models

Best for

Data engineers performing impact analysis before model changes

Teams building data governance dashboards that need lineage visualization

LLM agents that need to understand data dependencies for code generation or documentation

Requires

dbt-core 1.0+ with manifest.json artifact

Python 3.8+ for graph traversal logic

MCP client with tool-calling capability

Limitations

Lineage limited to dbt-managed models and sources — does not include external systems or non-dbt transformations

Requires manifest.json to be up-to-date; stale manifests produce inaccurate lineage

Does not track runtime data flow or actual column-level transformations, only declared dependencies

What makes it unique

Constructs lineage graphs directly from manifest.json node relationships without requiring dbt execution, enabling instant dependency queries. Supports bidirectional traversal (upstream sources and downstream consumers) with explicit relationship typing (depends_on, ref, source).

vs alternatives

Faster than dbt Cloud's lineage API for local projects because it operates on local artifacts, and provides more detailed relationship metadata than simple dependency lists.

column-level lineage and data type tracking

Medium confidence

Extracts column-level lineage information from dbt manifest by parsing model contracts, column definitions, and test metadata. Maps columns through transformation chains to track data types, nullability, and documentation across upstream and downstream models. Implements column-to-column dependency tracking using manifest column metadata and test associations.

Solves for

Trace how a specific column transforms through dbt models from source to final tableIdentify which columns are tested and what tests apply to themVerify data type consistency and nullability constraints across model chainGenerate column-level documentation and lineage diagrams for data catalogs

Best for

Data quality teams building column-level lineage for compliance and governance

Teams implementing data contracts and schema validation in dbt

LLM agents generating documentation or SQL based on column-level dependencies

Requires

dbt-core 1.5+ with manifest.json containing column metadata

Models with defined contracts or explicit column documentation

Python 3.8+ for column graph construction

Limitations

Column lineage limited to columns explicitly defined in dbt model contracts or tests — inferred columns not tracked

Does not track column transformations at SQL expression level; only model-to-model column mapping

Requires dbt 1.5+ with column-level contracts enabled for full accuracy

What makes it unique

Extracts column-level lineage from dbt manifest contracts and test metadata, enabling fine-grained tracking of data transformations. Combines column definitions, test associations, and data type information into unified lineage graph without requiring SQL parsing.

vs alternatives

Provides column-level detail that simple model lineage cannot offer, and requires no external data catalog or SQL parsing — all information comes from dbt artifacts.

dbt documentation content retrieval and search

Medium confidence

Indexes and retrieves dbt documentation content from manifest.json including model descriptions, column documentation, test descriptions, and source definitions. Exposes documentation as searchable text content accessible via MCP tools, enabling LLM agents to cite and reference dbt documentation in responses. Implements text extraction from manifest metadata fields without requiring dbt docs server.

Solves for

Retrieve documentation for a specific model, column, or source to include in LLM responsesSearch dbt documentation by keyword or pattern to find related modelsAccess test descriptions and expectations for data quality contextBuild documentation-aware LLM agents that cite dbt docs when answering data questions

Best for

Teams using Claude or other LLM clients to ask questions about dbt models

Building AI-powered data documentation assistants

Automating documentation generation or validation workflows

Requires

dbt-core 1.0+ with manifest.json containing documentation metadata

Models and columns with documentation strings defined in dbt project

Python 3.8+ for text extraction and indexing

Limitations

Documentation limited to text content in manifest — does not include dbt docs website HTML or custom styling

Search is basic text matching; no semantic search or embedding-based retrieval

Markdown formatting in documentation may not be fully preserved in output

What makes it unique

Extracts and indexes dbt documentation directly from manifest.json without requiring dbt docs server, making documentation accessible to LLM agents via MCP. Treats dbt docs as structured knowledge base queryable by model, column, or test.

vs alternatives

Enables documentation retrieval without running dbt docs server, and integrates documentation directly into LLM context — faster and more seamless than requiring agents to browse dbt docs website.

test coverage analysis and test metadata exposure

Medium confidence

Parses dbt test definitions from manifest.json and maps tests to models and columns they validate. Exposes test metadata including test type (generic/singular), test parameters, and expected outcomes. Enables analysis of test coverage gaps by identifying untested models and columns. Implements test-to-model mapping using manifest test node relationships.

Solves for

Identify which models and columns have test coverage and which are untestedRetrieve test definitions and parameters for a specific modelAnalyze test coverage gaps and recommend tests for critical modelsGenerate test coverage reports for data quality governance

Best for

Data quality teams implementing test coverage standards

Teams automating test recommendations for new models

Building data governance dashboards with test coverage metrics

Requires

dbt-core 1.0+ with manifest.json containing test definitions

Tests defined in dbt project (tests/ directory or inline)

Python 3.8+ for test metadata parsing

Limitations

Only captures tests defined in dbt project — external data quality tools not included

Test execution results not available; only test definitions from manifest

Does not track test failure history or flakiness

What makes it unique

Maps test definitions to models and columns via manifest relationships, enabling coverage analysis without executing tests. Treats test metadata as queryable knowledge base for data quality governance.

vs alternatives

Provides test coverage insights without running dbt test, and integrates test metadata into LLM context for intelligent test recommendations.

source and external table metadata exposure

Medium confidence

Extracts source definitions from manifest.json including source names, table names, database/schema locations, and source-level documentation. Exposes source metadata as queryable information enabling LLM agents to understand raw data inputs and their properties. Implements source node parsing from manifest with support for source freshness checks and source-level tests.

Solves for

Retrieve source table definitions and their database locationsUnderstand data freshness expectations and source-level testsMap sources to downstream models that depend on themGenerate documentation about raw data inputs and their quality expectations

Best for

Teams building data lineage tools that need source context

LLM agents answering questions about raw data inputs

Data governance teams tracking external data dependencies

Requires

dbt-core 1.0+ with manifest.json containing source definitions

Sources defined in sources.yml or equivalent dbt project files

Python 3.8+ for source metadata extraction

Limitations

Source metadata limited to dbt sources.yml definitions — does not connect to actual database schemas

Freshness checks defined in dbt but not executed; only metadata exposed

Does not validate that source tables actually exist in database

What makes it unique

Exposes dbt source definitions from manifest as queryable metadata, enabling LLM agents to understand raw data inputs and their properties without querying actual databases.

vs alternatives

Provides source context without database connections, making it lightweight and fast for lineage and documentation use cases.

mcp protocol integration and tool exposure

Medium confidence

Implements MCP (Model Context Protocol) server that exposes dbt metadata capabilities as standardized tools callable by MCP-compatible clients (Claude, Cline, etc.). Uses MCP server framework to define tool schemas, handle client requests, and return structured responses. Enables seamless integration of dbt metadata into LLM agent workflows through standard MCP tool-calling interface.

Solves for

Make dbt metadata queryable from Claude or other MCP clients via natural languageEnable LLM agents to call dbt metadata tools as part of reasoning chainsIntegrate dbt context into multi-step workflows that combine dbt queries with other toolsExpose dbt capabilities to non-technical users through LLM interface

Best for

Teams using Claude Desktop or Cline with MCP support

Building LLM agents that need dbt context in multi-tool workflows

Non-technical users querying dbt metadata through natural language

Requires

Python 3.8+ with MCP server library (e.g., mcp package)

MCP-compatible client (Claude Desktop 0.1+, Cline, etc.)

dbt-core 1.0+ with manifest.json artifact

Limitations

Requires MCP-compatible client; does not work with standard REST APIs or direct Python imports

Tool schemas must be pre-defined; dynamic tool generation not supported

Response size limited by MCP protocol and client constraints

What makes it unique

Implements full MCP server wrapping dbt metadata capabilities, enabling seamless tool-calling from Claude and other MCP clients. Uses standard MCP protocol for schema definition and request/response handling.

vs alternatives

Provides native MCP integration that works out-of-box with Claude Desktop and Cline, versus requiring custom API wrappers or Python SDK imports.

manifest.json parsing and caching

Medium confidence

Reads and parses dbt manifest.json artifact into in-memory data structures for fast metadata queries. Implements caching of parsed manifest to avoid repeated file I/O and JSON deserialization. Handles manifest schema variations across dbt versions and provides error handling for missing or corrupted manifests. Uses Python JSON parsing with optional caching layer for performance.

Solves for

Load dbt project metadata once and serve multiple queries efficientlyHandle manifest updates when dbt parse is re-runSupport multiple dbt projects with separate manifest instancesProvide fast metadata access for interactive LLM agent workflows

Best for

Interactive workflows requiring sub-second metadata queries

Long-running LLM agent sessions that query dbt metadata repeatedly

Multi-project environments with separate manifests

Requires

dbt-core 1.0+ with manifest.json artifact

Python 3.8+ with json module

Sufficient memory for manifest size (typically <100MB for most projects)

Limitations

Manifest must be manually refreshed by running dbt parse; no automatic detection of changes

Large manifests (1000+ models) may consume significant memory

No built-in manifest versioning or rollback capability

What makes it unique

Implements efficient manifest parsing with optional caching layer, enabling fast repeated queries without re-parsing JSON. Handles manifest schema variations across dbt versions.

vs alternatives

Faster than repeatedly executing dbt commands or parsing manifest on each query, and more flexible than dbt Cloud API for local projects.

Capabilities are decomposed by AI analysis. Each maps to specific user intents and improves with match feedback.

Related Artifactssharing capabilities

Artifacts that share capabilities with dbt-docs, ranked by overlap. Discovered automatically through the match graph.

MCP Server25

dbt

** - Official MCP server for [dbt (data build tool)](https://www.getdbt.com/product/what-is-dbt) providing integration with dbt Core/Cloud CLI, project metadata discovery, model information, and semantic layer querying capabilities.

dbt project metadata discovery and graph traversaldbt language server protocol (lsp) integration for column-level lineage

2 shared capabilities

Product27

Euno

Transforms data modeling with seamless dbt™ integration and...

automated lineage documentation and dependency mappingdbt documentation generation and enrichment

2 shared capabilities

Platform44

Elementary

Open-source dbt-native data observability and anomaly detection.

schema change detection and lineage trackingdbt-native anomaly detection via statistical test generation

2 shared capabilities

Product29

Metaplane

Monitor, manage, and enhance data integrity...

dbt-transformation-monitoringdata-lineage-visualization

2 shared capabilities

MCP Server44

OpenMetadata

OpenMetadata is a unified metadata platform for data discovery, data observability, and data governance powered by a central metadata repository, in-depth column level lineage, and seamless team collaboration.

column-level lineage tracking and visualization

1 shared capability

Platform40

Monte Carlo

Enterprise data observability with ML-powered anomaly detection.

data lineage tracking and visualization

1 shared capability

Best For

✓dbt-core OSS users building LLM agents that need project awareness
✓Teams integrating dbt metadata into Claude or other MCP-compatible clients
✓Data engineers automating dbt project analysis and governance
✓Data engineers performing impact analysis before model changes
✓Teams building data governance dashboards that need lineage visualization
✓LLM agents that need to understand data dependencies for code generation or documentation
✓Data quality teams building column-level lineage for compliance and governance
✓Teams implementing data contracts and schema validation in dbt

Known Limitations

⚠Requires pre-compiled manifest.json — does not execute dbt commands, so metadata reflects last successful dbt parse/run
⚠No real-time project state — changes to dbt_project.yml require manual dbt parse to update manifest
⚠Limited to dbt-core projects; does not support dbt Cloud-specific features like discovery API
⚠Lineage limited to dbt-managed models and sources — does not include external systems or non-dbt transformations
⚠Requires manifest.json to be up-to-date; stale manifests produce inaccurate lineage
⚠Does not track runtime data flow or actual column-level transformations, only declared dependencies

Requirements

dbt-core 1.0+ with compiled manifest.json artifactPython 3.8+ for MCP server runtimeMCP-compatible client (Claude Desktop, Cline, etc.)dbt-core 1.0+ with manifest.json artifactPython 3.8+ for graph traversal logicMCP client with tool-calling capabilitydbt-core 1.5+ with manifest.json containing column metadataModels with defined contracts or explicit column documentation

Input / Output

Accepts: file paths (dbt_project.yml, manifest.json), project directory structure, model name or source name (string), direction parameter (upstream/downstream), model name and column name (string), optional filter for test type or data type, model/column/source name (string), search query (string), optional filter for documentation type, model name or column name (string), optional test type filter (generic/singular), source name and table name (string), optional filter for source properties, MCP tool calls with JSON parameters, natural language queries (processed by LLM client), file path to manifest.json, optional cache configuration parameters

Produces: JSON structured metadata, text summaries of project configuration, JSON graph structure with nodes and edges, list of related models with relationship types, JSON column lineage graph with data types and nullability, list of tests applied to column, column documentation and descriptions, documentation text content, JSON with documentation metadata and source references, list of matching models/columns with relevance, JSON test metadata with parameters and definitions, test coverage statistics (count of tested vs untested models), list of tests applied to specific model/column, JSON source metadata with database/schema/table info, source documentation and freshness expectations, list of models depending on source, JSON structured responses conforming to MCP protocol, text summaries for LLM consumption, parsed manifest data structures, cache statistics and metadata

UnfragileRank

Adoption15%(30% weight)

Quality25%(25% weight)

Ecosystem30%(25% weight)

Match Graph10%(15% weight)

Freshness75%(5% weight)

UnfragileRank is computed from adoption signals, documentation quality, ecosystem connectivity, match graph feedback, and freshness. No artifact can pay for a higher rank.

Type: MCP Server

8 capabilities

Visit dbt-docs→

About

** - MCP server for dbt-core (OSS) users as the official dbt MCP only supports dbt Cloud. Supports project metadata, model and column-level lineage and dbt documentation.

Alternatives to dbt-docs

IntelliCode50Extension

AI-assisted development

Compare →

GitHub Copilot Chat53Extension

AI chat features powered by Copilot

Compare →

GitHub Copilot52Extension

Your AI pair programmer

Compare →

Claude Code for VS Code52Extension

Claude Code for VS Code: Harness the power of Claude Code without leaving your IDE

Compare →

Are you the builder of dbt-docs?

Claim this artifact to get a verified badge, access match analytics, see which intents users search for, and manage your listing.

Claim this artifact →Verification via email

Get the weekly brief

New tools, rising stars, and what's actually worth your time. No spam.

Data Sources

github awesome

Looking for something else?

Search →

Capabilities8 decomposed

dbt project metadata extraction and exposure

Medium confidence

Solves for

Best for

dbt-core OSS users building LLM agents that need project awareness

Teams integrating dbt metadata into Claude or other MCP-compatible clients

Data engineers automating dbt project analysis and governance

Requires

dbt-core 1.0+ with compiled manifest.json artifact

Python 3.8+ for MCP server runtime

MCP-compatible client (Claude Desktop, Cline, etc.)

Limitations

Requires pre-compiled manifest.json — does not execute dbt commands, so metadata reflects last successful dbt parse/run

No real-time project state — changes to dbt_project.yml require manual dbt parse to update manifest

Limited to dbt-core projects; does not support dbt Cloud-specific features like discovery API

What makes it unique

vs alternatives

Faster and lighter than dbt Cloud MCP for local dbt-core projects because it reads cached artifacts instead of making API calls, and requires no dbt Cloud subscription.

model-level lineage graph construction and traversal

Medium confidence

Solves for

Best for

Data engineers performing impact analysis before model changes

Teams building data governance dashboards that need lineage visualization

LLM agents that need to understand data dependencies for code generation or documentation

Requires

dbt-core 1.0+ with manifest.json artifact

Python 3.8+ for graph traversal logic

MCP client with tool-calling capability

Limitations

Lineage limited to dbt-managed models and sources — does not include external systems or non-dbt transformations

Requires manifest.json to be up-to-date; stale manifests produce inaccurate lineage

Does not track runtime data flow or actual column-level transformations, only declared dependencies

What makes it unique

vs alternatives

Faster than dbt Cloud's lineage API for local projects because it operates on local artifacts, and provides more detailed relationship metadata than simple dependency lists.

column-level lineage and data type tracking

Medium confidence

Solves for

Best for

Data quality teams building column-level lineage for compliance and governance

Teams implementing data contracts and schema validation in dbt

LLM agents generating documentation or SQL based on column-level dependencies

Requires

dbt-core 1.5+ with manifest.json containing column metadata

Models with defined contracts or explicit column documentation

Python 3.8+ for column graph construction

Limitations

Column lineage limited to columns explicitly defined in dbt model contracts or tests — inferred columns not tracked

Does not track column transformations at SQL expression level; only model-to-model column mapping

Requires dbt 1.5+ with column-level contracts enabled for full accuracy

What makes it unique

vs alternatives

Provides column-level detail that simple model lineage cannot offer, and requires no external data catalog or SQL parsing — all information comes from dbt artifacts.

dbt documentation content retrieval and search

Medium confidence

Solves for

Best for

Teams using Claude or other LLM clients to ask questions about dbt models

Building AI-powered data documentation assistants

Automating documentation generation or validation workflows

Requires

dbt-core 1.0+ with manifest.json containing documentation metadata

Models and columns with documentation strings defined in dbt project

Python 3.8+ for text extraction and indexing

Limitations

Documentation limited to text content in manifest — does not include dbt docs website HTML or custom styling

Search is basic text matching; no semantic search or embedding-based retrieval

Markdown formatting in documentation may not be fully preserved in output

What makes it unique

vs alternatives

Enables documentation retrieval without running dbt docs server, and integrates documentation directly into LLM context — faster and more seamless than requiring agents to browse dbt docs website.

test coverage analysis and test metadata exposure

Medium confidence

Solves for

Best for

Data quality teams implementing test coverage standards

Teams automating test recommendations for new models

Building data governance dashboards with test coverage metrics

Requires

dbt-core 1.0+ with manifest.json containing test definitions

Tests defined in dbt project (tests/ directory or inline)

Python 3.8+ for test metadata parsing

Limitations

Only captures tests defined in dbt project — external data quality tools not included

Test execution results not available; only test definitions from manifest

Does not track test failure history or flakiness

What makes it unique

vs alternatives

Provides test coverage insights without running dbt test, and integrates test metadata into LLM context for intelligent test recommendations.

source and external table metadata exposure

Medium confidence

Solves for

Best for

Teams building data lineage tools that need source context

LLM agents answering questions about raw data inputs

Data governance teams tracking external data dependencies

Requires

dbt-core 1.0+ with manifest.json containing source definitions

Sources defined in sources.yml or equivalent dbt project files

Python 3.8+ for source metadata extraction

Limitations

Source metadata limited to dbt sources.yml definitions — does not connect to actual database schemas

Freshness checks defined in dbt but not executed; only metadata exposed

Does not validate that source tables actually exist in database

What makes it unique

Exposes dbt source definitions from manifest as queryable metadata, enabling LLM agents to understand raw data inputs and their properties without querying actual databases.

vs alternatives

Provides source context without database connections, making it lightweight and fast for lineage and documentation use cases.

mcp protocol integration and tool exposure

Medium confidence

Solves for

Best for

Teams using Claude Desktop or Cline with MCP support

Building LLM agents that need dbt context in multi-tool workflows

Non-technical users querying dbt metadata through natural language

Requires

Python 3.8+ with MCP server library (e.g., mcp package)

MCP-compatible client (Claude Desktop 0.1+, Cline, etc.)

dbt-core 1.0+ with manifest.json artifact

Limitations

Requires MCP-compatible client; does not work with standard REST APIs or direct Python imports

Tool schemas must be pre-defined; dynamic tool generation not supported

Response size limited by MCP protocol and client constraints

What makes it unique

vs alternatives

Provides native MCP integration that works out-of-box with Claude Desktop and Cline, versus requiring custom API wrappers or Python SDK imports.

manifest.json parsing and caching

Medium confidence

Solves for

Best for

Interactive workflows requiring sub-second metadata queries

Long-running LLM agent sessions that query dbt metadata repeatedly

Multi-project environments with separate manifests

Requires

dbt-core 1.0+ with manifest.json artifact

Python 3.8+ with json module

Sufficient memory for manifest size (typically <100MB for most projects)

Limitations

Manifest must be manually refreshed by running dbt parse; no automatic detection of changes

Large manifests (1000+ models) may consume significant memory

No built-in manifest versioning or rollback capability

What makes it unique

Implements efficient manifest parsing with optional caching layer, enabling fast repeated queries without re-parsing JSON. Handles manifest schema variations across dbt versions.

vs alternatives

Faster than repeatedly executing dbt commands or parsing manifest on each query, and more flexible than dbt Cloud API for local projects.

Capabilities are decomposed by AI analysis. Each maps to specific user intents and improves with match feedback.

Alternatives to dbt-docs

IntelliCode50Extension

AI-assisted development

Compare →

GitHub Copilot Chat53Extension

AI chat features powered by Copilot

Compare →

GitHub Copilot52Extension

Your AI pair programmer

Compare →

Claude Code for VS Code52Extension

Claude Code for VS Code: Harness the power of Claude Code without leaving your IDE

Compare →

dbt-docs

Capabilities8 decomposed

dbt project metadata extraction and exposure

model-level lineage graph construction and traversal

column-level lineage and data type tracking

dbt documentation content retrieval and search

test coverage analysis and test metadata exposure

source and external table metadata exposure

mcp protocol integration and tool exposure

manifest.json parsing and caching

Related Artifactssharing capabilities

dbt

Euno

Elementary

Metaplane

OpenMetadata

Monte Carlo

Best For

Known Limitations

Requirements

Input / Output

UnfragileRank

About

Categories

Alternatives to dbt-docs

Are you the builder of dbt-docs?

Get the weekly brief

Data Sources

dbt-docs

Capabilities8 decomposed

dbt project metadata extraction and exposure

model-level lineage graph construction and traversal

column-level lineage and data type tracking

dbt documentation content retrieval and search

test coverage analysis and test metadata exposure

source and external table metadata exposure

mcp protocol integration and tool exposure

manifest.json parsing and caching

Related Artifactssharing capabilities

dbt

Euno

Elementary

Metaplane

OpenMetadata

Monte Carlo

Best For

Known Limitations

Requirements

Input / Output

UnfragileRank

About

Categories

Alternatives to dbt-docs

Are you the builder of dbt-docs?

Get the weekly brief

Data Sources