pbix-mcp

An MCP server for creating, reading, writing, and evaluating Power BI .pbix and .pbit files — no Power BI Desktop required. The entire PBIX binary format has been independently reversed and reimplemented in pure Python — no templates, no skeletons, no Microsoft binaries. Generated files open in PBI Desktop with full interactivity: view data, add measures, create visuals, and refresh — verified with PBI Desktop March 2026.

Exposes 112 tools covering report creation (all 6 data types, cross-table relationships, CSV/SQLite/SQL Server/MySQL/PostgreSQL/Excel/JSON/Azure SQL data sources, DirectQuery, and DAX measures), layout editing, visual management, bookmarks, custom visuals, custom HTML/CSS/SVG visuals (with report cross-filtering — see docs/html-visuals.md), service-portable rich content (certified AppSource visual references incl. Deneb, SVG data-URI image measures, Desktop-complete field parameters — see docs/rich-content.md), field parameters, calculation groups, TMDL export, incremental refresh, DAX evaluation (156 functions), RLS security, and binary format internals.

See CHANGELOG.md for version history.

Try It

Generate a complete 3-page Northwind Analytics Dashboard in under a second:

pip install pbix-mcp
python examples/create_showcase.py

Creates a PBIX with 6 tables, 5 relationships (including chained cross-table lookups), 4 DAX measures, 3 pages, and 14 visuals. Open showcase_northwind.pbix in Power BI Desktop — everything works: slicers, cross-filtering, drill-through.

Quick Start

git clone https://github.com/d0nk3yhm/pbix-mcp.git
cd pbix-mcp
pip install -e .

Claude Desktop / Claude Code

Add to your MCP config file:

{
  "mcpServers": {
    "powerbi-editor": {
      "command": "pbix-mcp-server"
    }
  }
}

Windows note: If pbix-mcp-server is not on PATH, use the full Python path:

{
  "mcpServers": {
    "powerbi-editor": {
      "command": "python",
      "args": ["-m", "pbix_mcp.cli"]
    }
  }
}

Codex Desktop (OpenAI)

1. pip install git+https://github.com/d0nk3yhm/pbix-mcp.git
2. Open Codex Desktop → Settings → MCP → Add Server
3. Configure:
   • Name: powerbi-editor
   • Command: pbix-mcp-server
   • Arguments: (leave empty)

Generic MCP (stdio)

pbix-mcp-server
# With debug logging:
pbix-mcp-server --log-level debug

Format Reversal Status

Every layer of the PBIX binary format has been independently reversed and reimplemented. No templates, skeletons, or Microsoft binaries are used.

The only non-generated artifact is the 144-byte CryptKey constant. This is a Microsoft RSA key BLOB that requires rskeymgmt infrastructure to generate. The key is GUID-independent — any valid key works with any database ID. Random bytes produce PFE_INVALID_CRYPT_KEY.

Stability

Known Limitations

• DAX engine is best-effort — designed for practical evaluation, not semantic parity with Analysis Services. Unsupported functions return None with status "unsupported", circular references raise DAXEvaluationError. See docs/supported-dax.md for full details.
• PBIR format — PBI Desktop (March 2026) has rendering bugs with PBIR decomposed format. PBIP export uses legacy report format (version 1.0) which works reliably.
• 1 out of 204 tested measures returns BLANK (requires per-employee RANKX visual row context)
• Performance — tables >100K rows trigger a warning; the DAX engine operates on in-memory Python data
• Opening existing DirectQuery files — layout, measures, and metadata editing work; DAX evaluation and table reads return clear errors since data lives in the remote source (this is inherent to DirectQuery — the data isn't in the file)
• Creating DirectQuery files — fully working with SQL Server (LocalDB), PostgreSQL 16, and MySQL 9.6 (via MariaDB adapter); requires a running database server and initial data snapshot
• CryptKey.bin — the 144-byte RSA key BLOB cannot be generated without Microsoft's crypto infrastructure (rskeymgmt). A known-valid GUID-independent constant is used.
• Embedded VertiPaq data — verified working with 11 tables, 72 columns, 13 relationships, 121K+ rows (Adventure Works DW 2020) and 6 tables, 36 columns, 5 relationships, 25 rows, 3 pages, 14 visuals (Northwind showcase)
• RLE encoding — disabled in the VertiPaq encoder (pure bitpack used). Slightly less space-efficient but correct
• Adding partitions to PBIX — pbix_add_partition is blocked for PBIX files (needs PartitionStorage in VertiPaq). Works for PBIP/TMDL export. Reading and removing existing partitions works.
• Full DataModel rebuild — set_table_data, update_table_rows, add/remove_relationship, remove_table trigger a full DataModel rebuild via the builder pipeline. Most other tools (add_measure, modify_measure, modify_column, set_rls_role, add_perspective, add_culture, add_translations, update_data_source, etc.) use a lightweight metadata-only path.

Tools (112)

Create & File Management (5)

pbix_create · pbix_open · pbix_save · pbix_close · pbix_list_open

Report Layout & Visuals (22)

pbix_add_visual · pbix_remove_visual · pbix_format_visual · pbix_set_visual_sort · pbix_get_pages · pbix_add_page · pbix_remove_page · pbix_get_page_visuals · pbix_get_visual_detail · pbix_get_visual_positions · pbix_set_visual_property · pbix_update_visual_json · pbix_get_layout_raw · pbix_set_layout_raw · pbix_get_filters · pbix_set_filters · pbix_get_default_filters · pbix_get_settings · pbix_set_settings · pbix_get_bookmarks · pbix_add_bookmark · pbix_remove_bookmark

DAX Engine (4)

pbix_evaluate_dax · pbix_evaluate_dax_per_dimension · pbix_evaluate_calculated_columns · pbix_clear_dax_cache

DataModel Read (16)

pbix_get_model_schema · pbix_get_model_measures · pbix_get_model_relationships · pbix_get_model_power_query · pbix_get_model_columns · pbix_get_table_data · pbix_list_tables · pbix_get_metadata · pbix_list_data_sources · pbix_update_data_source · pbix_export_table_csv · pbix_export_all_tables_csv · pbix_find_value · pbix_query_table · pbix_table_stats · pbix_data_diff

DataModel Write (22)

pbix_datamodel_query_metadata · pbix_datamodel_modify_metadata · pbix_datamodel_add_measure · pbix_datamodel_modify_measure · pbix_datamodel_set_measure_category · pbix_datamodel_remove_measure · pbix_datamodel_modify_column · pbix_datamodel_add_relationship · pbix_datamodel_remove_relationship · pbix_datamodel_remove_table · pbix_datamodel_decompress · pbix_datamodel_recompress · pbix_datamodel_replace_file · pbix_datamodel_extract_file · pbix_datamodel_list_abf_files · pbix_set_table_data · pbix_update_table_rows · pbix_datamodel_add_field_parameter · pbix_datamodel_add_calculation_group · pbix_export_tmdl · pbix_export_pbip · pbix_replace_value

Resources, Themes & Custom Visuals (18)

pbix_list_resources · pbix_add_image · pbix_set_image · pbix_register_resource · pbix_get_theme · pbix_set_theme · pbix_extract_colors · pbix_recolor · pbix_get_linguistic_schema · pbix_set_linguistic_schema · pbix_add_custom_visual · pbix_reference_public_visual · pbix_remove_custom_visual · pbix_add_html_visual · pbix_get_html_visual · pbix_set_html_visual · pbix_html_template · pbix_svg_measure

DataMashup (2)

pbix_get_m_code · pbix_set_m_code

Row-Level Security (3)

pbix_get_rls_roles · pbix_set_rls_role · pbix_evaluate_rls

Perspectives (3)

pbix_get_perspectives · pbix_add_perspective · pbix_remove_perspective

User Hierarchies (3)

pbix_get_hierarchies · pbix_add_hierarchy · pbix_remove_hierarchy

Cultures & Translations (4)

pbix_get_cultures · pbix_add_culture · pbix_add_translations · pbix_remove_culture

Partition Management (3)

pbix_get_partitions · pbix_add_partition · pbix_remove_partition

Incremental Refresh (2)

pbix_set_incremental_refresh · pbix_get_incremental_refresh

Diagnostics & Security (5)

pbix_doctor · pbix_document · pbix_diff · pbix_performance · pbix_get_password

Creating Reports

Build a complete multi-table PBIX with relationships and cross-table DAX — no Power BI Desktop needed:

from pbix_mcp.builder import PBIXBuilder

builder = PBIXBuilder()

# Dimension table
builder.add_table('Products', [
    {'name': 'ProductID', 'data_type': 'Int64'},
    {'name': 'Product',   'data_type': 'String'},
    {'name': 'UnitPrice', 'data_type': 'Double'},
], rows=[
    {'ProductID': 1, 'Product': 'Widget A',    'UnitPrice': 29.99},
    {'ProductID': 2, 'Product': 'Widget B',    'UnitPrice': 49.99},
    {'ProductID': 3, 'Product': 'Gadget X',    'UnitPrice': 14.99},
])

# Fact table
builder.add_table('Sales', [
    {'name': 'OrderID',   'data_type': 'Int64'},
    {'name': 'ProductID', 'data_type': 'Int64'},
    {'name': 'Qty',       'data_type': 'Int64'},
    {'name': 'Region',    'data_type': 'String'},
], rows=[
    {'OrderID': 1001, 'ProductID': 1, 'Qty': 5,  'Region': 'North'},
    {'OrderID': 1002, 'ProductID': 2, 'Qty': 3,  'Region': 'South'},
    {'OrderID': 1003, 'ProductID': 3, 'Qty': 20, 'Region': 'East'},
])

# Cross-table relationship (from=many, to=one)
builder.add_relationship('Sales', 'ProductID', 'Products', 'ProductID')

# Measures (including cross-table RELATED)
builder.add_measure('Sales', 'Total Qty', 'SUM(Sales[Qty])')
builder.add_measure('Sales', 'Total Revenue',
    'SUMX(Sales, Sales[Qty] * RELATED(Products[UnitPrice]))')

builder.save('sales_report.pbix')

Opens in Power BI Desktop with full interactivity — slicers, cross-filtering, and all DAX measures work.

Refreshable CSV Sources

Point tables at external CSV files so data can be refreshed in Power BI Desktop:

builder.add_table('Sales', [
    {'name': 'OrderID',   'data_type': 'Int64'},
    {'name': 'ProductID', 'data_type': 'Int64'},
    {'name': 'Qty',       'data_type': 'Int64'},
], rows=sales_data,
   source_csv=r'C:\Data\sales.csv')  # M expression references this CSV

The initial data snapshot is embedded in the PBIX. When opened in Power BI Desktop, clicking Refresh re-imports from the CSV file. Edit the CSV → Refresh → data updates live.

Database Sources (SQL Server / SQLite / MySQL / PostgreSQL / Excel / JSON / Azure SQL)

Connect tables to databases so data can be refreshed from the DB:

# SQL Server (built-in PBI connector — works with LocalDB, Express, full)
builder.add_table('Orders', [
    {'name': 'OrderID', 'data_type': 'Int64'},
    {'name': 'Qty',     'data_type': 'Int64'},
], rows=orders_data,
   source_db={'type': 'sqlserver', 'server': r'(localdb)\MSSQLLocalDB',
              'database': 'MyDB', 'table': 'Orders'})

# SQLite (requires SQLite3 ODBC Driver — http://www.ch-werner.de/sqliteodbc/)
builder.add_table('Orders', [
    {'name': 'OrderID', 'data_type': 'Int64'},
    {'name': 'Qty',     'data_type': 'Int64'},
], rows=orders_data,
   source_db={'type': 'sqlite', 'path': r'C:\Data\mydb.sqlite', 'table': 'orders'})

# MySQL (built-in PBI connector — verified with MySQL 9.6)
builder.add_table('Orders', [
    {'name': 'OrderID', 'data_type': 'Int64'},
    {'name': 'Qty',     'data_type': 'Int64'},
], rows=orders_data,
   source_db={'type': 'mysql', 'server': 'localhost', 'database': 'mydb',
              'table': 'orders', 'port': 3306})

# PostgreSQL (built-in PBI connector — verified with PostgreSQL 16)
builder.add_table('Orders', [
    {'name': 'order_id', 'data_type': 'Int64'},
    {'name': 'qty',      'data_type': 'Int64'},
], rows=orders_data,
   source_db={'type': 'postgresql', 'server': 'localhost', 'database': 'mydb',
              'table': 'orders', 'port': 5432, 'schema': 'public'})

# MariaDB adapter (for MySQL DirectQuery — requires MariaDB ODBC 3.1 Driver)
builder.add_table('Orders', [
    {'name': 'OrderID', 'data_type': 'Int64'},
    {'name': 'Qty',     'data_type': 'Int64'},
], rows=orders_data,
   mode='directquery',
   source_db={'type': 'mariadb', 'server': 'localhost', 'database': 'mydb',
              'table': 'orders', 'port': 3306})

Data is Import mode by default — a snapshot is embedded in the PBIX at build time. Clicking Refresh in Power BI Desktop re-reads from the database. The report works offline between refreshes.

DirectQuery (Live Database Queries)

For true live queries (no refresh needed — data updates instantly):

builder.add_table('Orders', [
    {'name': 'OrderID', 'data_type': 'Int64'},
    {'name': 'Qty',     'data_type': 'Int64'},
], rows=snapshot_data,  # Initial snapshot (required)
   mode='directquery',
   source_db={'type': 'sqlserver', 'server': r'(localdb)\MSSQLLocalDB',
              'database': 'MyDB', 'table': 'Orders'})

DirectQuery creates a PBIX with Partition.Mode=1 and a Sql.Database() M expression. Power BI Desktop queries the database live — INSERT/UPDATE/DELETE in the database is reflected instantly without clicking Refresh.

Note: DirectQuery requires a running database server. Verified with SQL Server (LocalDB), PostgreSQL 16, and MySQL 9.6 (via MariaDB adapter). All three also support Import mode with Refresh. The rows parameter provides an initial data snapshot embedded in the PBIX.

Via MCP Tool

{
  "tool": "pbix_create",
  "arguments": {
    "file_path": "report.pbix",
    "tables_json": "[{\"name\": \"Sales\", \"columns\": [{\"name\": \"Amount\", \"data_type\": \"Double\"}], \"rows\": [{\"Amount\": 100}], \"source_csv\": \"C:/Data/sales.csv\"}]",
    "measures_json": "[{\"table\": \"Sales\", \"name\": \"Total\", \"expression\": \"SUM(Sales[Amount])\"}]",
    "relationships_json": "[{\"from_table\": \"Sales\", \"from_column\": \"ProductID\", \"to_table\": \"Products\", \"to_column\": \"ProductID\"}]"
  }
}

Switching Data Sources (No Rebuild)

Change connection strings on existing PBIX files without regenerating the DataModel — lightweight metadata-only update:

from pbix_mcp.formats.datamodel_roundtrip import decompress_datamodel, compress_datamodel
from pbix_mcp.formats.abf_rebuild import read_metadata_sqlite, rebuild_abf_with_modified_sqlite
from pbix_mcp.builder import _build_m_expression
import zipfile, io

# Open existing PBIX
with open('report.pbix', 'rb') as f:
    original = f.read()

z = zipfile.ZipFile(io.BytesIO(original))
abf = decompress_datamodel(z.read('DataModel'))

# Switch Sales table from SQL Server to PostgreSQL DirectQuery
def switch_source(conn):
    conn.row_factory = __import__('sqlite3').Row
    row = conn.execute(
        "SELECT p.ID, t.ID as tid FROM Partition p "
        "JOIN [Table] t ON p.TableID = t.ID WHERE t.Name = 'Sales'"
    ).fetchone()
    cols = [{'name': c['ExplicitName'],
             'data_type': {6:'Int64', 8:'Double', 2:'String'}[c['ExplicitDataType']]}
            for c in conn.execute(
                'SELECT ExplicitName, ExplicitDataType FROM [Column] '
                'WHERE TableID = ? AND Type = 1', (row['tid'],))]
    new_m = _build_m_expression('Sales', cols, source_db={
        'type': 'postgresql', 'server': 'pg.example.com', 'port': 5432,
        'database': 'analytics', 'table': 'sales', 'schema': 'public',
    }, is_directquery=True)
    conn.execute('UPDATE Partition SET QueryDefinition=?, Mode=1 WHERE ID=?',
                 (new_m, row['ID']))
    conn.commit()

new_abf = rebuild_abf_with_modified_sqlite(abf, switch_source)
new_dm = compress_datamodel(new_abf)

# Write back — only DataModel changes, rest of PBIX untouched
buf = io.BytesIO()
with zipfile.ZipFile(buf, 'w', zipfile.ZIP_DEFLATED) as z_out:
    with zipfile.ZipFile(io.BytesIO(original)) as z_in:
        for item in z_in.infolist():
            if item.filename == 'DataModel':
                z_out.writestr(item.filename, new_dm, compress_type=zipfile.ZIP_STORED)
            else:
                z_out.writestr(item, z_in.read(item.filename))
with open('report.pbix', 'wb') as f:
    f.write(buf.getvalue())

Supports all source types: sqlserver, postgresql, mysql, mariadb, sqlite, csv, excel, json, azuresql. Set is_directquery=True and Mode=1 for DirectQuery, or is_directquery=False and Mode=0 for Import.

Via MCP (Claude / Codex)

Just ask in plain English:

"List all data sources in my report"

"Switch the Sales table from SQL Server to PostgreSQL on pg.example.com, database analytics, DirectQuery"

"Change the CSV path for Products to C:\Data\new_products.csv"

The AI reads the current connections via pbix_list_data_sources, then calls pbix_update_data_source with the right parameters. No rebuild — only the connection metadata is updated.

Supported Data Types

VertiPaq Binary Format

Every component of the VertiPaq columnar storage engine is independently implemented:

• IDF — Bit-packed encoding for data columns (RLE disabled; pure bitpack is slightly less space-efficient but correct)
• IDFMETA — Segment statistics with tagged CP/CS/SS/SDOs blocks
• Dictionary — Type-specific encoding (Long/Real/String) with hash tables; large string dictionaries use canonical-Huffman compression (MS-XLDM §2.7.4, read + write) via xmhuffman (MIT)
• H$ system tables — Attribute hierarchy POS_TO_ID + ID_TO_POS using NoSplit<32> encoding
• R$ system tables — Relationship join INDEX using NoSplit encoding; +3 DATA_ID_OFFSET padding, 1-based row indices into TO table (derived from PBI Desktop ground truth binary comparison)
• Compression class IDs — Determined through binary format analysis (u32_a/u32_b selectors)
• XPress9 — Custom implementation of Power BI's DataModel compression format: reversed chunk framing, header signatures, single-thread and multi-thread container formats. The core XPress9 algorithm uses xpress9-python as a primitive; the full read/write/modify pipeline is original work
• ABF — Full archive generation: STREAM_STORAGE_SIGNATURE, BackupLogHeader, VirtualDirectory, BackupLog XML, data file layout

DAX Engine

156 functions across 10 categories. This is a best-effort evaluator — it produces correct results for common patterns but does not aim for semantic parity with Analysis Services.

Accuracy

Tested against 4 real-world Power BI dashboards (204 measures total). All 4 dashboards are publicly available from Dashboard-Design/Power-BI-Design-Files (MIT License, Sajjad Ahmadi). Anyone can download them and reproduce these results.

The 1 BLANK measure requires per-employee RANKX visual row context that doesn't exist at report level.

Verified Against Power BI Desktop

Safety

• pbix_save creates automatic .bak backups before overwriting
• pbix_close refuses to discard unsaved changes unless force=True
• SecurityBindings are auto-removed on repack (prevents corruption)
• All write operations are applied to temp directories, not directly to the original file

Testing

# Fast tests (no PBIX files needed, runs from fresh clone)
pytest -m "not slow"

# Download public test corpus (4 dashboards, MIT licensed)
python scripts/download_test_corpus.py --output-dir test_corpus

# Run integration tests against the corpus
PBIX_TEST_SAMPLES=test_corpus pytest tests/test_cross_report.py -v

From a fresh clone: 489 tests collected, 457 passed, 32 skipped, 0 failures. All 32 skipped tests are gated on the public test corpus (no private files are needed). Download it with python scripts/download_test_corpus.py, then set PBIX_TEST_SAMPLES=test_corpus to run them.

Architecture

PBIX file (ZIP)
├── Version                ← "1.28" UTF-16-LE (8 bytes)
├── [Content_Types].xml    ← OOXML package manifest
├── DiagramLayout          ← JSON: model diagram state
├── Settings               ← JSON: report settings
├── Metadata               ← JSON: file metadata
├── Report/Layout          ← JSON: pages, visuals, filters, data bindings
└── DataModel              ← XPress9 compressed → ABF archive
    ├── BackupLogHeader    ← XML: VDir offset, data offset, file count
    ├── ADDITIONAL_LOG     ← UTF-16: product name
    ├── PARTITIONS         ← UTF-16: partition marker
    ├── db.xml             ← XMLA Load document (28 namespaces)
    ├── CryptKey.bin       ← 144-byte RSA key BLOB (constant)
    ├── metadata.sqlitedb  ← SQLite: 68 system tables (Table, Column, Measure, Relationship, ...)
    ├── *.tbl\*.prt\*.idf  ← VertiPaq: bit-packed column data
    ├── *.idfmeta          ← Segment statistics (CP/CS/SS/SDOs)
    ├── *.dictionary       ← Dictionary encoding (Long/Real/String + hash)
    ├── H$*.tbl\...        ← Attribute hierarchy tables (NoSplit<32>)
    ├── R$*.tbl\...        ← Relationship index tables (NoSplit<N>)
    ├── BackupLog          ← XML: FileGroups, file paths, storage mappings
    └── VirtualDirectory   ← XML: file offsets and sizes

Package Layout

src/pbix_mcp/
  server.py              # MCP server (112 tools)
  cli.py                 # Entry point (pbix-mcp-server --log-level debug)
  builder.py             # PBIX builder (metadata, VertiPaq, layout, relationships)
  html_templates.py      # HTML/SVG template builders (kpi_card, bar_chart, gauge, table, …)
  svg_measures.py        # DAX codegen for SVG data-URI image measures (data_bar, sparkline, …)
  assets/pbix_html_visual/  # bundled "PBIX HTML" custom visual (.pbiviz) + source
  builder_v2.py          # Template-free ABF + ZIP generation
  errors.py              # Typed exceptions with stable error codes
  logging_config.py      # Diagnostic logging (normal/debug/trace)
  dax/
    engine.py            # DAX evaluator (156 functions, best-effort)
    calc_tables.py       # Calculated table support
  formats/
    abf_rebuild.py       # ABF archive reader and rebuilder
    datamodel_roundtrip.py  # XPress9 compress/decompress
    metadata_schema.py   # SQLite metadata schema (68 tables)
    model_reader.py      # Native VertiPaq table data reader (replaces PBIXRay)
    vertipaq_decoder.py  # VertiPaq IDF/dictionary/HIDX decoder
    vertipaq_encoder.py  # VertiPaq column encoding + NoSplit<N> encoder
  models/
    responses.py         # Pydantic response models
    requests.py          # Pydantic request models

Development

git clone https://github.com/d0nk3yhm/pbix-mcp.git
cd pbix-mcp
pip install -e ".[dev]"
pytest -m "not slow"
ruff check src/ tests/
mypy src/pbix_mcp/

See CONTRIBUTING.md for project conventions, SUPPORT.md for what counts as a bug vs unsupported behavior, and examples/ for runnable sample scripts.

Examples

Roadmap

• TMDL import — import models from TMDL files (export already implemented)
• Composite models — mixed Import + DirectQuery tables in the same report
• PBIR layout write — write reports in the new PBIR format alongside legacy

Architecture Notes

Incremental vs Full Rebuild

The builder generates the entire DataModel each time — metadata SQLite, VertiPaq column data, ABF container, and XPress9 compression. All offsets, checksums, and cross-references are computed from first principles.

For modifying existing PBIX files (adding a measure, changing a visual), the MCP server operates differently: it opens the file, modifies the specific layer (SQLite metadata for measures, JSON for layout), and repacks — without touching the VertiPaq binary data. This is true incremental editing.

No Microsoft Dependencies

This project is 100% Python with zero Microsoft DLLs or SDKs. Every layer of the PBIX format — from the ZIP shell to the VertiPaq column encoding — is independently reversed and implemented. The XPress9 compression uses xpress9-python (MIT) and the canonical-Huffman string store uses xmhuffman (MIT) as low-level primitives; the Power BI DataModel container format (chunk framing, headers, multi-thread support, full read/write/modify round-trip) is original work in datamodel_roundtrip.py.

Purpose & Interoperability

This project is an independent reimplementation of the Power BI .pbix file format, created for the purpose of interoperability — enabling AI agents, automation tools, and non-Windows platforms to create, read, and write Power BI files.

• No Microsoft source code was used. All binary format knowledge was derived through independent analysis of file structures and publicly observable behavior.
• Interoperability rights: In both the EU (Directive 2009/24/EC, Article 6) and US (DMCA 1201(f)), reverse engineering for interoperability purposes is a protected right that supersedes contractual restrictions.
• Functional specification: The binary format documentation in docs/vertipaq-spec.md describes functional information (data layouts, compression formats, metadata schemas) necessary for cross-platform compatibility.

This project is not affiliated with, endorsed by, or associated with Microsoft Corporation. "Power BI" and "PBIX" are trademarks of Microsoft.

License

MIT