arca-ai-vault 0.1.0

Python SDK for Arca - Private data vault for personal AI assistants

Arca Python SDK

Python SDK for Arca - A private data vault for personal AI assistants.

Overview

Arca provides two main APIs:

• Tables API: Store structured data with SQL-like queries (powered by DuckDB and Parquet)
• Vectors API: Store and search unstructured data semantically (powered by vector embeddings)

Installation

pip install arca-ai-vault

Or install from GitHub:

pip install git+https://github.com/xbora/arca-python-sdk.git

Or install from source:

git clone https://github.com/xbora/arca-python-sdk.git
cd arca-python-sdk
pip install -e .

Quick Start

Authentication

Get your API key from https://arca.build/api-keys

from arca import ArcaTableClient, ArcaVectorClient

# Initialize clients
table_client = ArcaTableClient(user_id="your-api-key")
vector_client = ArcaVectorClient(user_id="your-api-key")

Tables API - Store Structured Data

from arca import ArcaTableClient, TableColumn, SkillMetadata

client = ArcaTableClient(user_id="your-api-key")

# Create a table and insert data
response = client.upsert(
    table_name="meals",
    columns=[
        TableColumn("food", "VARCHAR"),
        TableColumn("calories", "INTEGER"),
        TableColumn("protein", "DOUBLE")
    ],
    data={
        "food": "Grilled Chicken",
        "calories": 165,
        "protein": 31.0
    },
    skill=SkillMetadata(
        description="Tracks daily meals and nutrition",
        examples=["SELECT * FROM meals WHERE calories > 200"]
    )
)

# Query data
results = client.query(
    table_name="meals",
    filters={"daysAgo": 7},
    order_by="calories DESC",
    limit=10
)

# List all tables
tables = client.list_tables()

# Get table schemas
schemas = client.get_schemas()

Vectors API - Semantic Search

from arca import ArcaVectorClient, VectorSkillMetadata, MetadataField

client = ArcaVectorClient(user_id="your-api-key")

# Add entries with automatic embedding generation
response = client.add(
    table_name="journal_entries",
    text="Today was incredibly productive. Finished the big project ahead of schedule.",
    metadata={
        "category": "personal",
        "mood": "positive",
        "date": "2024-01-15"
    },
    skill=VectorSkillMetadata(
        description="Personal journal entries with mood tracking",
        metadata_fields=[
            MetadataField("category", "string", "Entry category", ["personal", "work", "health"]),
            MetadataField("mood", "string", "Emotional state", ["positive", "neutral", "negative"])
        ],
        search_examples=["Find days when I felt accomplished"],
        filter_examples=["category = 'personal' AND mood = 'positive'"]
    )
)

# Search semantically
results = client.search(
    table_name="journal_entries",
    query="productive and successful days",
    limit=5,
    filter="category = 'personal'"
)

# List all vector tables
tables = client.list_tables()

API Reference

ArcaTableClient

__init__(user_id: str, base_url: str = "https://arca.build")

Initialize the table client with your API key.

upsert(table_name, data, columns=None, skill=None)

Create or append to a table. This is the recommended method for inserting data.

Parameters:

• table_name (str): Name of the table
• data (dict): Dictionary of data to insert
• columns (list[TableColumn], optional): Column definitions (required on first insert)
• skill (SkillMetadata, optional): Metadata to help AI understand the table

query(table_name, query=None, filters=None, limit=None, offset=None, order_by=None, select=None, group_by=None, having=None)

Query a table with filters and aggregations.

Parameters:

• table_name (str): Name of the table to query
• query (str, optional): Raw SQL WHERE clause
• filters (dict, optional): Dictionary of filters (e.g., {"daysAgo": 7})
• limit (int, optional): Maximum number of results
• offset (int, optional): Number of results to skip
• order_by (str, optional): Column to order by (e.g., "created_at DESC")
• select (list[str], optional): List of columns to select
• group_by (str, optional): Column to group by
• having (str, optional): HAVING clause for aggregations

update(table_name, updates, where=None, filters=None)

Update rows in a table.

delete(table_name)

Delete an entire table.

list_tables()

List all tables for the authenticated user.

get_schemas()

Get schemas for all tables.

get_skill(table_name)

Get the SKILL.md file for a specific table.

update_skill(table_name, skill)

Update the SKILL.md file for a table.

get_all_skills()

Get all table skills in one request.

export(table_name)

Export a table as a Parquet file (returns bytes).

ArcaVectorClient

__init__(user_id: str, base_url: str = "https://arca.build")

Initialize the vector client with your API key.

add(table_name, text, metadata=None, generate_embedding=True, embedding=None, skill=None)

Add a vector entry with automatic embedding generation.

Parameters:

• table_name (str): Name of the vector table
• text (str): Text content to embed and store
• metadata (dict, optional): Optional metadata dictionary
• generate_embedding (bool): Whether to auto-generate embedding (default: True)
• embedding (list[float], optional): Pre-computed embedding vector
• skill (VectorSkillMetadata, optional): Metadata to help AI understand the table

search(table_name, query, limit=5, generate_embedding=True, embedding=None, filter=None)

Search vectors semantically.

Parameters:

• table_name (str): Name of the vector table to search
• query (str): Search query text
• limit (int): Maximum number of results (default: 5)
• generate_embedding (bool): Whether to auto-generate query embedding (default: True)
• embedding (list[float], optional): Pre-computed query embedding vector
• filter (str, optional): SQL-like filter expression (e.g., "category = 'personal'")

delete(table_name)

Delete an entire vector table.

list_tables()

List all vector tables for the authenticated user.

get_skill(table_name)

Get the SKILL.md file for a specific vector table.

update_skill(table_name, skill)

Update the SKILL.md file for a vector table.

get_all_skills()

Get all vector skills in one request.

export(table_name)

Export a vector table as a CSV file (returns bytes).

Data Models

TableColumn

Represents a column in an Arca table.

TableColumn(name: str, type: str, nullable: bool = True)

Supported types: VARCHAR, INTEGER, BIGINT, DOUBLE, BOOLEAN, TIMESTAMP, DATE, JSON

SkillMetadata

Metadata for table skills - helps AI understand how to use the table.

SkillMetadata(
    description: str = None,
    examples: list[str] = None,
    relationships: list[str] = None,
    notes: str = None
)

VectorSkillMetadata

Metadata for vector skills - helps AI understand how to search the vector table.

VectorSkillMetadata(
    description: str = None,
    metadata_fields: list[MetadataField] = None,
    search_examples: list[str] = None,
    filter_examples: list[str] = None,
    notes: str = None
)

MetadataField

Metadata field definition for vector tables.

MetadataField(
    name: str,
    type: str,
    description: str,
    examples: list[str] = None
)

Error Handling

The SDK provides custom exceptions for different error scenarios:

from arca import ArcaAPIError, ArcaAuthError, ArcaValidationError

try:
    response = client.query(table_name="meals")
except ArcaAuthError as e:
    print(f"Authentication failed: {e}")
except ArcaAPIError as e:
    print(f"API error: {e.message} (status: {e.status_code})")
except ArcaValidationError as e:
    print(f"Validation error: {e}")

Examples

See the examples/ directory for more comprehensive examples:

• table_examples.py - Tables API usage examples
• vector_examples.py - Vectors API usage examples

Development

Install development dependencies:

pip install -e ".[dev]"

Run tests:

pytest

License

MIT License

Links

• Documentation
• API Reference
• GitHub Repository
• Arca Homepage