Context Hub (Chub) Documentation

The demo showcases several core LiveKit functionalities:

• Access Token Generation: Create JWT tokens with configurable user identity, display name, room permissions (join, publish, subscribe), and expiration.
• Room Management: Use the server API to create, list, and delete rooms, as well as manage participants.
• Recording (Egress): Start room composite recordings and configure output formats.
• Real-time Connection: Connect to rooms, handle track subscriptions, and process video/audio streams.
• RPC Methods: Register RPC method handlers and perform remote procedure calls between participants.

Use this package for control-plane operations (managing the service itself).

Use this for:

• Creating, deleting, or updating domains.
• Inspecting domain status, endpoints, or configuration.
• Changing instance sizing, storage, security, or logging settings.
• Managing tags, packages, upgrades, and software updates.

DO NOT use this for:

• Indexing documents.
• Running searches (/_search).
• Bulk operations (/_bulk).

For data-plane traffic (searching/indexing), use DescribeElasticsearchDomainCommand to fetch the domain endpoint, then use a signed HTTP client to send requests to that endpoint.

The google-cloud-memcache package is an Admin API client. It is used for provisioning, inspecting, and managing Memcached instances.

Do NOT use this package for Memcached key/value traffic (get, set, delete, etc.).

Correct Workflow:

1. Use google-cloud-memcache to create/inspect the instance.
2. Retrieve the discovery_endpoint or node addresses from the Instance object.
3. Use a standard Memcached protocol client (e.g., pymemcache) in your application to perform data operations using those endpoints.

Select a primitive based on your security requirements:

Choose between crawl and map based on your specific use case:

Use Crawl when:

• You need full content extraction.
• You are building RAG systems.
• You need to process paginated or nested content.

Use Map when:

• You need quick site structure discovery.
• You only need to collect URLs without content.
• You are planning a crawl strategy.

Select a client based on your integration requirements:

• WebClient: For standard Slack Web API calls.
• AsyncWebClient: For asynchronous code (requires aiohttp).
• WebhookClient: For incoming webhooks and response_url usage.
• SocketModeClient: For WebSocket-based app connections (requires an app-level xapp- token).

Context Hub uses a multi-layered scoring model to rank search results. The final score is calculated using the following formula:

final_score = term_relevance * (1 + qualityScore / 20) * source_boost

1. Term Relevance (Match)

Uses weighted field scoring (BM25-style) instead of flat points:

• id: 3.0 weight
• name: 2.5 weight
• tags: 2.0 weight
• description: 1.0 weight

2. Description Quality Score

A deterministic score (0-10) computed at build time and stored as _qualityScore. It acts as a multiplier: a perfect score (10) provides a 1.5x boost.

3. Source Authority Boost

A multiplier based on the entry's source field:

• maintainer: 1.3x
• official: 1.2x
• community: 1.0x

Select a type based on your requirements for key uniqueness, case sensitivity, and mutability:

• MultiDict: Mutable mapping that allows duplicate keys and preserves insertion order. Use this for URL query parameters or form fields.
• CIMultiDict: Like MultiDict, but key lookup is case-insensitive. Ideal for HTTP headers.
• MultiDictProxy: A read-only, live dynamic view over a MultiDict.
• CIMultiDictProxy: A read-only, live dynamic view over a CIMultiDict.
• istr: A str subclass used for case-insensitive key handling.

Note: Keys must be str instances or subclasses of str (like istr).

To manage Azure networking resources (VNets, subnets, NSGs, etc.) via the management plane, install azure-mgmt-network along with azure-identity for authentication. It is recommended to pin the version to 30.2.0 for compatibility with this guide.

Using pip:

python -m pip install "azure-mgmt-network==30.2.0" azure-identity

Using uv:

uv add "azure-mgmt-network==30.2.0" azure-identity

Using poetry:

poetry add "azure-mgmt-network==30.2.0" azure-identity

Document AI is regional. You must initialize the DocumentProcessorServiceClient with a client_options object where the api_endpoint matches the location of your processor (e.g., us or eu).

Required parameters:

• project_id
• location (e.g., us, eu)
• processor_id
• mime_type (e.g., application/pdf, image/png)

Failure to match the endpoint to the processor location will result in errors.

from google.api_core.client_options import ClientOptions
from google.cloud import documentai

location = "us"

client = documentai.DocumentProcessorServiceClient(
    client_options=ClientOptions(
        api_endpoint=f"{location}-documentai.googleapis.com"
    )
)