README.md

See Docs/API/Redis.md for the practical API surface and usage examples.

Package structure

• Configuration for provider options
• DependencyInjection for service registration
• Keys for Redis key conventions
• Storage for the public store facade
• Storage/Candidates for Redis index candidate selection
• Storage/Candidates/Models for candidate plan models
• Storage/Candidates/Planning for candidate plan construction
• Storage/Candidates/Execution for candidate plan execution
• Storage/Documents for request document loading
• Storage/Documents/Keys for Redis document key creation
• Storage/Documents/Payloads for bulk payload reads
• Storage/Documents/Materialization for request document deserialization and ordering
• Storage/Documents/Reading for document read orchestration
• Storage/Identifiers for Redis set and id reads
• Storage/Identifiers/Models for identifier set operation models
• Storage/Identifiers/Loading for Redis id set loading and normalization
• Storage/Persistence for write-side request persistence
• Storage/Persistence/Models for persistence record models
• Storage/Persistence/Reading for single request persistence reads
• Storage/Persistence/Writing for write transactions, payload creation, and index maintenance
• Storage/Queries for query orchestration and result materialization
• Storage/Queries/Reading for request query orchestration
• Storage/Queries/Materialization for query result shaping
• Serialization for request payload handling
• Serialization/Converters for custom JSON converters

Usage

using Microsoft.Extensions.DependencyInjection;
using ModularityKit.Mutator.Governance.Redis;
using StackExchange.Redis;

var services = new ServiceCollection();
var multiplexer = await ConnectionMultiplexer.ConnectAsync("localhost:6379");

services.AddRedisGovernanceStore(
    multiplexer,
    options => options.KeyPrefix = "modularitykit:governance");

Current query strategy

The provider uses Redis indexes first and then applies the storage-agnostic governance query evaluator in memory for the final filter pass.

Today this means:

• point reads and optimistic concurrency stay fully Redis-backed
• common queue views are narrowed by Redis set membership
• broad ad hoc filters still finish in memory after candidate selection