Tag: pattern

Dynamically select which tools an LLM agent can use based on the task context. Reduce token usage and improve decision quality by narrowing the tool set.

Cache LLM embeddings to reduce API calls and cost. Store embeddings with a content hash key and serve from cache on repeated inputs.

Pause LLM agent execution for human approval before high-impact actions. Route decisions to a reviewer when confidence is low or stakes are high.

Fall back to alternative LLM providers or models when the primary fails. Handle rate limits, timeouts, and errors gracefully with a provider chain.

Validate LLM inputs and outputs with rules, classifiers, and content filters. Prevent prompt injection, toxic content, and data leakage before reaching users.

Route queries to different LLM models based on complexity, cost, and latency requirements. Classify input before dispatching to the right model.

Chain multiple LLM calls where each step's output feeds the next step's input. Break complex tasks into smaller, verifiable prompts for better results.

Combine keyword (BM25) and semantic (vector) search to improve retrieval accuracy in RAG pipelines. Fuse ranked results using reciprocal rank fusion.

How to offload cross-cutting concerns to a proxy ambassador. Covers connection pooling, retry logic, circuit breaking, monitoring, and TLS termination for client services.

How to isolate legacy systems with translation adapters. Covers ACL facade, domain translation, bidirectional mapping, and gradual legacy replacement.

How to create dedicated backends per client type. Covers BFF for web, mobile, and desktop. Covers API aggregation, client-specific optimization, and GraphQL BFF.

Combine workloads into fewer compute resources to reduce cost, improve utilization, and simplify operations.

Centralize application configuration outside of deployment artifacts to support live updates and multi-environment management.

Route requests to multiple backend services through a single entry point that handles cross-cutting concerns.

Expose lightweight health endpoints so orchestrators, load balancers, and monitoring tools can verify service availability.

Coordinate a single active instance among multiple distributed nodes to avoid conflicts and split-brain scenarios.

How to build a modular monolith with strict internal module boundaries. Covers module isolation, shared kernel, inter-module communication, and migration to microservices.

Isolate tenant data in shared infrastructure using row-level security, schema-per-tenant, or database-per-tenant strategies. A pattern for SaaS applications.

Chain processing steps with independent filters connected by pipes. A pattern for data transformation pipelines where each step is reusable and composable.

How to extend services with companion containers for cross-cutting concerns. Covers sidecar containers, shared volumes, health probes, and service mesh sidecars.

How to gradually replace a legacy system by intercepting routes and routing traffic to new services. Covers strangler fig, incremental migration, and cutover.

Delegate authentication to external identity providers. A pattern for integrating OAuth2, OIDC, SAML, and SSO across multiple services and organizations.

Validate claims and delegate access using signed vouchers without exposing sensitive data. A security pattern for token-based authorization between services.

How to bridge batch and streaming pipelines with a data lake. Covers Lambda architecture, Kafka Connect S3 sink, schema alignment, and unified serving layer.

How to stream database changes to downstream consumers with CDC. Covers log-based CDC, Debezium, Kafka Connect, outbox pattern, and consumer reconciliation.

How to track data origin and transformations end-to-end. Covers column-level lineage, OpenLineage, Marquez, metadata injection, and impact analysis.

How to build ETL pipelines with extract, transform, and load stages. Covers staging tables, incremental extraction, idempotent loads, and orchestration.

How to re-run data loads safely without duplicates. Covers deduplication keys, MERGE upserts, load IDs, partition overwrite, and transactional loads.

How to manage schema versions for streaming pipelines with a schema registry. Covers Avro, backward compatibility, forward compatibility, and consumer migration.

Create families of related objects without specifying concrete classes. A creational design pattern for consistent object families.

Wrap a database table or view in a class where an instance is tied to a single row, and the class provides methods for CRUD operations directly on the object.

Isolate state in actors that communicate only via messages. Each actor processes one message at a time, eliminating shared-state concurrency bugs by design.

Convert the interface of a class into another interface clients expect. A structural design pattern for interface compatibility.

Encapsulate a cluster of domain objects treated as a single unit for data changes. An Aggregate Root controls access to its internal entities and value objects.

Stream data lazily with async generators. Yield values one at a time as they become available, enabling memory-efficient processing of large or infinite data sequences.

Prevent upstream systems from overwhelming downstream consumers by propagating flow-control signals backward through the pipeline, ensuring stable throughput under load.

Create dedicated backend services tailored to the specific needs of each frontend client type, aggregating downstream APIs and optimizing data shapes per platform.

A shared knowledge space where independent specialized modules collaborate to solve complex problems by contributing partial solutions.

Run two identical environments and switch traffic between them. Deploy to the idle environment, test it, then flip the router for instant release or rollback.

Decouple an abstraction from its implementation so both can vary independently. A structural design pattern for platform independence.

Construct complex objects step by step. A creational design pattern for readable, configurable object construction.

Reduce coupling between presentation and business tiers by introducing an intermediary that handles lookup, creation, and invocation of business services.

Load data into the cache on demand from the backing store. A caching pattern that gives the application full control over what and when to cache.

Strategies for keeping cached data fresh: TTL expiration, explicit invalidation, write-through, and event-driven cache eviction.

Prevent thundering herd cache misses with locks, single-flight, and early refresh strategies to protect the database from concurrent reloads.

Route a small percentage of traffic to the new version while the rest stays on stable. Monitor health metrics and gradually increase or roll back based on results.

Pass requests along a chain of handlers until one handles it. A behavioral design pattern for decoupling senders and receivers.

Prevent cascading failures by stopping requests to failing services. An architectural pattern for resilient distributed systems.

Store large payloads in external storage and pass only a lightweight reference token through the message bus, reducing broker load and preventing message size limits from being exceeded.

Encapsulate a request as an object, letting you parameterize clients with queues, logs, and undoable operations. A behavioral design pattern.

Undo the effects of a completed transaction by executing a counter-operation, enabling eventual consistency in long-running business processes across distributed services.

Map a coarse-grained entity to multiple database tables by composing dependent objects, reducing the number of fine-grained remote calls in EJB and distributed systems.

Compose objects into tree structures to represent part-whole hierarchies. A structural design pattern for treating individual objects and compositions uniformly.

Distribute static and live content through geographically dispersed edge servers to reduce latency, improve availability, and offload origin infrastructure.

Encapsulate state and services needed by multiple components into a single context object, reducing method signature bloat and decoupling code from specific environment details.

Separate read and write operations into different models, optimizing each for their specific workload. A data pattern for scalable systems.

Abstract and encapsulate all access to a data source by exposing a clean interface while hiding persistence details from business logic.

Separate in-memory domain objects from the database by delegating persistence to a dedicated mapper layer, keeping models framework-agnostic.

Give each microservice its own private database to ensure loose coupling, independent deployment, and technology heterogeneity across the application portfolio.

Route unprocessable messages to a separate dead letter queue for inspection and replay. Prevent poison messages from blocking the main queue indefinitely.

Add new functionality to objects dynamically by wrapping them. A structural design pattern for flexible behavior extension.

Supply dependencies from outside rather than creating them internally. An architectural pattern for decoupled, testable code.

Roll out changes progressively in rings of increasing size. Start with a small group, verify health, then expand to larger rings before full deployment.

Coordinate mutually exclusive access to shared resources across distributed nodes using a consensus-based lock service, preventing race conditions in scaled-out systems.

Capture and publish major occurrences within a domain model to decouple side effects from core business logic and enable reactive workflows.

Load related data in a single query rather than multiple round-trips, preventing the N+1 problem and improving read performance.

Compose entities from pure data components and process them with systems, enabling high-performance and flexible game object architecture without deep inheritance.

Decouple components by routing events through a central bus. A behavioral pattern for loosely coupled communication between modules.

Replicate state changes across services by publishing events that carry the full updated entity state, enabling consumers to maintain their own local copies without querying the source.

Store the state of an application as a sequence of events rather than storing only the current state. An architectural pattern for audit-friendly systems.

Provide a simplified interface to a complex subsystem. A structural pattern that hides implementation details behind a clean API.

Create objects without specifying the exact class to instantiate. A creational design pattern for flexible object creation.

Share objects to support large numbers of fine-grained objects efficiently. A structural design pattern for memory optimization.

Route all incoming requests through a single handler that dispatches to the appropriate page command, centralizing request processing and security.

Place a validation and security boundary at the edge of a system to inspect, sanitize, and authenticate all incoming requests before they reach internal services.

Distribute data across nodes with partitioning so each node owns a shard. Horizontal scaling without shared state, with locality and fault isolation per partition.

Degrade functionality instead of failing when dependencies are unavailable. Serve partial results, cached data, or fallback features to keep users running.

Process messages from a queue exactly once regardless of duplicates by using idempotent operations, unique identifiers, and deduplication strategies at the consumer level.

Ensure each object is loaded only once per transaction by caching instances by their primary key, preventing duplicate in-memory representations of the same database row.

Use a dedicated inbox table or queue to record incoming events or requests, ensuring reliable delivery, deduplication, and idempotent processing even when downstream systems fail.

Compose cross-cutting concerns into a chain of pluggable filters that intercept requests and responses, enabling reusable preprocessing and postprocessing logic.

Define a representation for a language's grammar along with an interpreter that uses the representation to interpret sentences. A behavioral design pattern for mini-languages.

Provide a way to access elements of a collection sequentially without exposing its underlying representation. A behavioral design pattern for traversal.

Build high-throughput queues using atomic operations instead of locks. Multiple threads can enqueue and dequeue concurrently without blocking or context-switching overhead.

Encapsulate lifecycle, coordination, and access control for a set of related objects through a dedicated manager class that centralizes operations and enforces invariants.

Use empty interfaces as metadata tags to signal properties or capabilities at compile time and runtime, enabling type-safe checks without modifying class behavior.

Precompute and store expensive query results in a read-optimized cache to avoid repeated costly aggregation or joins across large datasets.

Define an object that encapsulates how a set of objects interact. A behavioral design pattern for reducing chaotic dependencies.

Capture and restore an object's internal state without violating encapsulation. A behavioral design pattern for undo/redo.

Prevent duplicate processing by tracking message IDs with idempotency keys. Consumers check a store before processing to skip messages already handled.

Delay message processing to a scheduled time. Move messages that cannot be processed now to a deferred queue or schedule them for later delivery.

Smooth traffic spikes by placing a queue between a producer and a consumer. The producer writes messages at any rate; the consumer processes them at a steady pace.

Add reusable behavior to classes without inheritance by composing methods from shared objects into a target class.

Separate presentation logic from the view by introducing a presenter that intermediates between the model and a passive view, enabling testable UI code.

Bind UI components declaratively to a ViewModel that exposes data and commands, enabling automatic synchronization between view and state.

Encapsulate private state and behavior inside a self-contained unit with a public API. A structural pattern for organizing code into reusable, scope-safe modules.

Manage a map of named singleton instances, providing controlled access to a finite set of shared objects identified by keys.

Separate application into Model, View, and Controller components. An architectural design pattern for organized, maintainable code.

Use a default object instead of null references to eliminate null checks and simplify client code. A behavioral pattern for safer defaults.

Reuse expensive objects instead of creating and destroying them repeatedly. A creational pattern for managing scarce resources efficiently.

Define a subscription mechanism to notify multiple objects about events. A behavioral design pattern for event-driven communication.

Reliably publish domain events by persisting them in an outbox table within the same database transaction as the business operation.

Use a dedicated controller object for each logical page in a web application, handling the request and populating the view for that specific page.

Split a single class definition across multiple source files to separate auto-generated code from hand-written code, or to organize large classes by concern.

Enable third-party extensions by defining extension points in a host system that loads and executes live external modules.

Process tasks based on priority rather than arrival order, ensuring high-priority work gets resources before lower-priority tasks even if it arrived later.

Decouple production and consumption with a shared queue. Producers generate items at their own pace; consumers process them independently through a bounded or unbounded buffer.

Create new objects by copying existing ones. A creational design pattern for cloning and object duplication.

Provide a surrogate or placeholder for another object to control access to it. A structural design pattern for access control, lazy loading, and logging.

Broadcast events to multiple independent subscribers. Publishers send messages to a topic without knowing which subscribers exist, enabling loose coupling between producers and consumers.

Introduce a queue between task producers and consumers to smooth out traffic spikes, decouple components, and prevent downstream services from being overwhelmed by burst workloads.

Process asynchronous data streams with backpressure. Subscribers request N items at a time, preventing fast producers from overwhelming slow consumers.

A transparent cache layer that intercepts read requests, fetches from the data source on miss, and populates the cache automatically.

Proactively refresh cache entries before they expire to eliminate cache misses on hot keys and maintain consistent read latency.

Centralize access to shared services and objects via a lookup table. A structural pattern that decouples consumers from concrete implementations.

Abstract data access logic behind a clean interface. An architectural design pattern for testable, maintainable data layers.

Retry an operation that has failed with transient errors, using configurable strategies like fixed delay, exponential backoff, or circuit breaker integration.

Assign dynamic roles to objects at runtime instead of hard-coding behavior in class hierarchies, enabling flexible identity changes without inheritance bloat.

Manage distributed transactions across multiple services by chaining local transactions with compensating actions for rollbacks. A microservices pattern.

Coordinate resilient job scheduling by separating scheduling logic from execution agents and adding a supervisor that monitors, restarts, and manages agent lifecycle.

Preserve message ordering for related messages in a distributed system by grouping them into ordered sequences and processing them one at a time through a single consumer.

Manage database connections across serverless invocations by using external connection poolers, connection reuse, and lightweight clients to avoid connection exhaustion.

Store function state as an append-only event log so workflows can be replayed, audited, and recovered without a persistent database.

Broadcast a single event to multiple independent consumers via SNS, EventBridge, or SQS so each consumer processes the event without coupling.

Chain serverless functions via Step Functions or orchestration layers to build multi-step workflows with retries, branching, and state management.

Handle backpressure in serverless by using SQS, token buckets, and concurrency limits to protect downstream services from burst traffic.

Keep Lambda functions warm by sending periodic ping events to reduce cold start latency for latency-sensitive workloads.

Split a large dataset into smaller partitions (shards) distributed across multiple servers to improve growth, performance, and availability beyond single-node limits.

Drop requests proactively under extreme load to protect the system. Reject excess traffic before it consumes resources and causes cascading failures.

Ensure a class has only one instance and provide global access to it. A creational design pattern for controlled object creation.

Encapsulate business rules for selecting objects as reusable, composable predicate objects that can be combined with logical operators.

Allow an object to alter its behavior when its internal state changes. A behavioral design pattern for finite state machines.

Deploy static files to a dedicated content delivery network or object storage to offload origin servers, reduce latency, and improve availability for assets like images, CSS, and JavaScript.

Define a family of algorithms, encapsulate each one, and make them interchangeable. A behavioral design pattern for flexible behavior selection.

Define the skeleton of an algorithm in a base class, letting subclasses override specific steps without changing the algorithm's structure. A behavioral design pattern.

Reuse a fixed set of threads for short-lived tasks instead of creating a new thread per task. Reduces overhead and bounds resource usage under load.

Limit the rate at which a system processes requests or consumes resources to prevent overload, ensure fair usage, and maintain predictable performance under varying load.

Prevent operations from hanging indefinitely by enforcing a maximum execution time. A resilience pattern for predictable response times.

Provide an alternative to multiple inheritance by linking two separate classes through mutual references, allowing them to delegate methods to each other as needed.

Combine an L1 in-memory cache with an L2 distributed cache to reduce latency for hot keys while maintaining cache consistency across instances.

Define game object types as runtime data rather than hard-coding them as classes, enabling designers to create new entity variants without recompiling the codebase.

Track changes to in-memory objects during a business transaction and commit all updates atomically to the database, ensuring consistency.

Model domain concepts by value rather than identity. An immutable object defined by its attributes, not by a unique ID.

Represent an operation to be performed on elements of an object structure without changing the classes of the elements. A behavioral design pattern.

Write to cache synchronously and persist to the database asynchronously for high-throughput write workloads with eventual consistency.

Synchronously write to both cache and backing store so the cache always has the latest data without TTL-based invalidation.

How to separate data-fetching logic from rendering in React using the container-presenter pattern. Covers hooks migration, testing benefits, and trade-offs.

How to organize CSS with utility-first classes and component-scoped layers. Covers Tailwind CSS, CSS layers, BEM, CSS modules, and design tokens.

How to compose reusable logic with custom React hooks. Covers hook composition patterns, dependency arrays, context integration, and testing strategies.

How to ship interactivity only where needed using islands architecture. Covers Astro islands, partial hydration, React islands, and performance benefits.

How to update UI immediately and reconcile on server response in React. Covers rollback on error, conflict resolution, and React Query integration.

How to build a functional HTML baseline and progressively enhance with JavaScript. Covers core functionality, feature detection, graceful degradation, and accessibility.

How to model UI state transitions with finite state machines in React. Covers XState, statecharts, guarded transitions, and preventing impossible states.

How to use React Suspense boundaries for declarative loading states. Covers data fetching, streaming SSR, nested boundaries, and error boundaries.

Resolve nested GraphQL fields in a single batched request to eliminate N+1 queries and reduce database load.

Implement Relay-style cursor-based pagination with edges, nodes, and pageInfo for stable GraphQL list queries.

Coalesce individual load requests into batched calls with per-request caching to prevent N+1 queries and redundant fetches.

Attach structured metadata to GraphQL errors using extension codes for predictable client-side error handling.

Share entity types across federated GraphQL services so the gateway can resolve fields from multiple subgraphs transparently.

Model polymorphic types with GraphQL interfaces to share field contracts across different object types while keeping resolvers type-specific.

Centralize input validation for GraphQL mutations using custom validators, schema directives, and structured error responses.

Merge multiple independent GraphQL schemas into a single unified schema that clients can query as one graph.

How to expose circuit breaker state as metrics for observability. Covers Prometheus integration, alerting rules, dashboards, and state transitions.

How to propagate correlation IDs across service boundaries for end-to-end request tracing. Covers HTTP headers, message queues, and logging integration.

How to propagate trace context across service boundaries with OpenTelemetry. Covers span creation, context propagation, sampling, and trace analysis.

How to implement liveness and readiness probes for container orchestration. Covers Kubernetes probes, dependency checks, graceful degradation, and probe endpoints.

How to collect, tag, and aggregate business metrics for observability. Covers Prometheus, OpenTelemetry, custom metrics, histograms, and dashboarding.

How to emit structured JSON logs with consistent fields for searchability. Covers Python structlog, Winston, Serilog, log levels, and log aggregation.

How to isolate resources per service to limit blast radius. Covers thread pool isolation, connection pool partitioning, semaphore-based bulkheads, and resource quotas.

How to test service recovery with half-open circuit breaker state transitions. Covers closed, open, half-open states, trial requests, and gradual recovery.

How to switch to a standby system on primary failure detection. Covers active-passive, active-active, health monitoring, DNS fallover, database replication, and automated promotion.

How to drain in-flight requests before process exit. Covers signal handling, health check removal, connection draining, timeout enforcement, and cleanup hooks.

How to implement token bucket rate limiting for API protection. Covers bucket refill, burst handling, per-key buckets, distributed rate limiting with Redis, and sliding windows.

How to retry failed operations with exponential backoff and random jitter. Covers full jitter, equal jitter, decorrelated jitter, retry budgets, and idempotency.

How to use contract testing to verify that API producers and consumers agree on request and response shapes. Covers Pact consumer-driven contracts and provider verification.

How to use setup and teardown fixtures to create reusable test context. Covers beforeEach, factory functions, fixture objects, and cleanup with examples.

How to use golden master testing to characterize legacy code behavior before refactoring. Covers capturing output, comparing baselines, and incremental refactoring.

How to use mock servers to isolate integration tests from external dependencies. Covers WireMock, nock, MSW, and Mountebank with configuration examples.

How to write parameterized tests to verify the same logic across multiple inputs. Covers pytest parametrize, Jest test.each, JUnit ParameterizedTest, and data providers.

How to use snapshot testing to detect unintended changes in serialized output. Covers Jest snapshots, pytest snapshot, and inline vs external snapshots.

How to use test doubles to isolate units under test. Covers stubs, spies, fakes, mocks, and dummy objects with examples in Python, JavaScript, and Java.

How to structure a test suite using the test pyramid. Covers unit, integration, and E2E test proportions, the testing trophy, and ice cream cone anti-pattern.