# The T-Shaped Dev

Production-grade engineering for full-stack JavaScript developers who want to think like architects -- React, Node, software architecture, and building real AI systems. No fluff, every week.

This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles.

## 6 Things a Senior Engineer Covers When a Payment Is Retried

DevFeed: [6 Things a Senior Engineer Covers When a Payment Is Retried](<https://devfeed.tech/articles/6-things-a-senior-engineer-covers-when-a-payment-is-retried-138255.md>)

Original publisher: [Read original article](<https://thetshaped.dev/p/how-to-design-a-payments-endpoint-idempotency>)

Author: The T-Shaped Dev

Published: 2026-11-17T11:03:34Z

Content type: article

Language: en

Sources: [The T-Shaped Dev](<https://devfeed.tech/sources/the-t-shaped-dev.md>)

Topics: [payment-gateway](<https://devfeed.tech/topics/payment-gateway.md>), [stripe](<https://devfeed.tech/topics/stripe.md>), [Query (disambiguation)](<https://devfeed.tech/topics/query.md>), [Finance](<https://devfeed.tech/topics/finance.md>)

Tags: [adyen](<https://devfeed.tech/tags/adyen.md>), [api](<https://devfeed.tech/tags/api.md>), [architecture](<https://devfeed.tech/tags/architecture.md>), [astro](<https://devfeed.tech/tags/astro.md>), [async](<https://devfeed.tech/tags/async.md>), [backends](<https://devfeed.tech/tags/backends.md>), [call](<https://devfeed.tech/tags/call.md>), [card](<https://devfeed.tech/tags/card.md>), [charged](<https://devfeed.tech/tags/charged.md>), [charges](<https://devfeed.tech/tags/charges.md>), [charging](<https://devfeed.tech/tags/charging.md>), [checkout](<https://devfeed.tech/tags/checkout.md>), [click](<https://devfeed.tech/tags/click.md>), [client](<https://devfeed.tech/tags/client.md>), [cloudways](<https://devfeed.tech/tags/cloudways.md>), [cloudways-velocity](<https://devfeed.tech/tags/cloudways-velocity.md>), [crash](<https://devfeed.tech/tags/crash.md>), [key](<https://devfeed.tech/tags/key.md>), [order](<https://devfeed.tech/tags/order.md>), [payment](<https://devfeed.tech/tags/payment.md>)

### AI overview

A retry-safe payment flow anchors each attempt to a server-created order. A database partial unique index permits one live attempt per order; `INSERT ... ON CONFLICT DO NOTHING` chooses the winner, avoiding the race in check-then-insert. Commit the processing row before calling the provider. Retries and lease takeovers reuse the attempt's stored amount, card, and provider key, so a charge completed before a server crash can be replayed safely; declines release the order for a new attempt. Unknown outcomes stay processing, and expired leases are handled by later requests or a sweeper. Provider keys have retention limits: the excerpt says Stripe keeps them at least 24 hours, so older attempts require reconciliation rather than another charge attempt. Queue receipts and payment events with the success update in an outbox, whose database guard prevents duplicate effects even when email or webhook delivery retries.

### Source excerpt

Where the key comes from, one payment per order, the decline, the crash, the receipt, and what the interviewer grades.

## I Knew the Terms. I Couldn't Explain Them.

DevFeed: [I Knew the Terms. I Couldn't Explain Them.](<https://devfeed.tech/articles/i-knew-the-terms-i-couldn-t-explain-them-130856.md>)

Original publisher: [Read original article](<https://thetshaped.dev/p/i-knew-the-terms-i-couldnt-explain>)

Author: Petar Ivanov

Published: 2026-11-12T11:01:59Z

Content type: article

Language: en

Sources: [The T-Shaped Dev](<https://devfeed.tech/sources/the-t-shaped-dev.md>)

Topics: [data type](<https://devfeed.tech/topics/data-type.md>), [await](<https://devfeed.tech/topics/await.md>), [architecture-components](<https://devfeed.tech/topics/architecture-components.md>)

Tags: [3](<https://devfeed.tech/tags/3.md>), [about](<https://devfeed.tech/tags/about.md>), [after](<https://devfeed.tech/tags/after.md>), [also](<https://devfeed.tech/tags/also.md>), [api](<https://devfeed.tech/tags/api.md>), [api-design](<https://devfeed.tech/tags/api-design.md>), [architecture](<https://devfeed.tech/tags/architecture.md>), [array](<https://devfeed.tech/tags/array.md>), [async](<https://devfeed.tech/tags/async.md>), [await](<https://devfeed.tech/tags/await.md>), [away](<https://devfeed.tech/tags/away.md>), [bug](<https://devfeed.tech/tags/bug.md>), [can](<https://devfeed.tech/tags/can.md>), [course](<https://devfeed.tech/tags/course.md>), [explain](<https://devfeed.tech/tags/explain.md>), [failed](<https://devfeed.tech/tags/failed.md>), [free](<https://devfeed.tech/tags/free.md>), [i](<https://devfeed.tech/tags/i.md>), [interview](<https://devfeed.tech/tags/interview.md>), [terms](<https://devfeed.tech/tags/terms.md>)

### AI overview

The course targets a gap between recognizing JavaScript concepts and explaining them aloud: learners answer each question before seeing a model answer. In its async example, `forEach` discards promises returned by async callbacks, so the loop finishes before the work; the excerpt also notes that async predicates passed to `filter` are a bug because promises are truthy. It offers `Promise.all(items.map(...))` to await parallel work or `for...of` for sequential work. The course is described as having 660+ questions and architecture scenarios across React, Node.js, TypeScript and system design, with answers tagged by level. The free sample includes three chapters, 76 questions, three architecture scenarios and chapter self-tests. These details come from the supplied excerpt.

### Source excerpt

The interview I almost failed, the course I built around it, and 3 free chapters you can try today.

## 5 LLM Output Failures That Still Get Past Structured Outputs

DevFeed: [5 LLM Output Failures That Still Get Past Structured Outputs](<https://devfeed.tech/articles/5-llm-output-failures-that-still-get-past-structured-outputs-126815.md>)

Original publisher: [Read original article](<https://thetshaped.dev/p/llm-structured-outputs-always-parse>)

Author: The T-Shaped Dev

Published: 2026-11-10T11:28:54Z

Content type: article

Language: en

Sources: [The T-Shaped Dev](<https://devfeed.tech/sources/the-t-shaped-dev.md>)

Topics: [Reliable AI outputs](<https://devfeed.tech/topics/reliable-ai-outputs.md>), [LLM observability](<https://devfeed.tech/topics/llm-observability.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>), [TypeScript](<https://devfeed.tech/topics/typescript.md>), [MSP MCP](<https://devfeed.tech/topics/msp-mcp.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [alert](<https://devfeed.tech/tags/alert.md>), [anthropic](<https://devfeed.tech/tags/anthropic.md>), [api](<https://devfeed.tech/tags/api.md>), [create](<https://devfeed.tech/tags/create.md>), [decoder](<https://devfeed.tech/tags/decoder.md>), [dr](<https://devfeed.tech/tags/dr.md>), [employee](<https://devfeed.tech/tags/employee.md>), [enum](<https://devfeed.tech/tags/enum.md>), [error](<https://devfeed.tech/tags/error.md>), [failures](<https://devfeed.tech/tags/failures.md>), [field](<https://devfeed.tech/tags/field.md>), [ids](<https://devfeed.tech/tags/ids.md>), [llm](<https://devfeed.tech/tags/llm.md>), [rules](<https://devfeed.tech/tags/rules.md>), [sdk](<https://devfeed.tech/tags/sdk.md>), [structured](<https://devfeed.tech/tags/structured.md>), [structured-outputs](<https://devfeed.tech/tags/structured-outputs.md>), [typescript](<https://devfeed.tech/tags/typescript.md>), [values](<https://devfeed.tech/tags/values.md>)

### AI overview

Structured outputs guarantee schema shape, not truth: plausible but incorrect values and invented IDs can still pass. The article says to inspect the schema actually sent, since in the described `@anthropic-ai/sdk` 1.6.0 TypeScript helper, rules such as positive numbers and enums may become unenforced description hints and are caught only by Zod parsing. Use `messages.create()` to check `stop_reason` first and route refusals or token cut-offs to review, then parse values and verify IDs against application data. Track field distributions and pass rates by prompt version to detect valid outputs drifting over time. The excerpt's examples and SDK behavior are specific to the versions and APIs it describes.

### Source excerpt

Legal but wrong values, IDs that point at nothing, rules your SDK never sends, refusals and drift. Where each check lives now, in TypeScript.

## Find Your Stage on the Way to JavaScript Architect

DevFeed: [Find Your Stage on the Way to JavaScript Architect](<https://devfeed.tech/articles/find-your-stage-on-the-way-to-javascript-architect-122184.md>)

Original publisher: [Read original article](<https://thetshaped.dev/p/find-your-stage-on-the-way-to-javascript>)

Author: Petar Ivanov

Published: 2026-11-07T14:30:57Z

Content type: tutorial

Language: en

Sources: [The T-Shaped Dev](<https://devfeed.tech/sources/the-t-shaped-dev.md>)

Topics: [roadmaps](<https://devfeed.tech/topics/roadmaps.md>), [Test-driven development](<https://devfeed.tech/topics/tdd.md>), [Publish-subscribe pattern](<https://devfeed.tech/topics/pubsub.md>)

Tags: [architect](<https://devfeed.tech/tags/architect.md>), [free](<https://devfeed.tech/tags/free.md>), [javascript](<https://devfeed.tech/tags/javascript.md>), [plan](<https://devfeed.tech/tags/plan.md>), [roadmap](<https://devfeed.tech/tags/roadmap.md>)

### AI overview

The roadmap's practical takeaway is to identify the first stage with an unchecked skill, then complete that stage's work project before moving on. Its six ordered stages progress from JavaScript fundamentals and maintainable code to application and database design, distributed systems, and architecture decisions spanning teams. The 20-item self-check asks whether you can perform and explain each skill today without looking it up; the author estimates two to four months per stage with five focused hours weekly, depending on work experience. The excerpt gives a Stage 4 exercise: use `pg_stat_statements` or a provider's query insights to inspect the three queries with the most total time, read their plans, improve one with an index or rewrite, and record before-and-after numbers. The full roadmap is described as a free seven-page PDF, delivered after email signup, which also subscribes the reader to the author's email list.

### Source excerpt

A 20-box self-check across six stages, and a free 7-page roadmap with the plan for each one.

## 5 Postgres Migrations That Work Fine in Dev and Lock Up Production

DevFeed: [5 Postgres Migrations That Work Fine in Dev and Lock Up Production](<https://devfeed.tech/articles/5-postgres-migrations-that-work-fine-in-dev-and-lock-up-production-117140.md>)

Original publisher: [Read original article](<https://thetshaped.dev/p/postgres-schema-migrations-lock-queue-zero-downtime-alter-table>)

Author: The T-Shaped Dev

Published: 2026-11-04T11:42:44Z

Content type: article

Language: en

Sources: [The T-Shaped Dev](<https://devfeed.tech/sources/the-t-shaped-dev.md>)

Topics: [Transactions](<https://devfeed.tech/topics/transactions.md>)

Tags: [api](<https://devfeed.tech/tags/api.md>), [batch](<https://devfeed.tech/tags/batch.md>), [concurrent](<https://devfeed.tech/tags/concurrent.md>), [concurrently](<https://devfeed.tech/tags/concurrently.md>), [constraint](<https://devfeed.tech/tags/constraint.md>), [database](<https://devfeed.tech/tags/database.md>), [deploy](<https://devfeed.tech/tags/deploy.md>), [downtime](<https://devfeed.tech/tags/downtime.md>), [lock](<https://devfeed.tech/tags/lock.md>), [migrations](<https://devfeed.tech/tags/migrations.md>), [postgres](<https://devfeed.tech/tags/postgres.md>), [production](<https://devfeed.tech/tags/production.md>)

### AI overview

Postgres migrations can stall production through lock queues even when the migration itself is fast: a waiting `ACCESS EXCLUSIVE` lock makes later queries wait behind it. Set `lock_timeout`, then retry failed deploy attempts; the cap limits but does not eliminate request stalls. For large tables, add `NOT VALID` constraints and validate them in a later deploy, and build indexes with `CREATE INDEX CONCURRENTLY` outside a transaction. Concurrent index builds are slower, wait for older transactions, and can leave invalid indexes that must be checked and removed or rebuilt. Keep rolling deploys compatible with old app instances by expanding, backfilling, then contracting across deploys. Run backfills as re-runnable, primary-key-batched jobs with commits between batches. Since Postgres 11, constant defaults can be added instantly; volatile defaults rewrite the table, while `now()` assigns existing rows the migration transaction's timestamp. The text notes that MySQL has a similar lock queue but requires different fixes.

### Source excerpt

Each one looks harmless until the table is big. Why it locks, what your users see, and the safe version of each.

## Domain Events Are Easy. Until One Fails Silently.

DevFeed: [Domain Events Are Easy. Until One Fails Silently.](<https://devfeed.tech/articles/domain-events-are-easy-until-one-fails-silently-95407.md>)

Original publisher: [Read original article](<https://thetshaped.dev/p/domain-events-production-failures-nodejs-typescript>)

Author: The T-Shaped Dev

Published: 2026-10-22T12:09:33Z

Content type: tutorial

Language: en

Sources: [The T-Shaped Dev](<https://devfeed.tech/sources/the-t-shaped-dev.md>)

Topics: [event driven](<https://devfeed.tech/topics/event-driven.md>), [Publish-subscribe pattern](<https://devfeed.tech/topics/pubsub.md>), [Transactions](<https://devfeed.tech/topics/transactions.md>), [await](<https://devfeed.tech/topics/await.md>)

Tags: [3](<https://devfeed.tech/tags/3.md>), [about](<https://devfeed.tech/tags/about.md>), [accounting](<https://devfeed.tech/tags/accounting.md>), [agents](<https://devfeed.tech/tags/agents.md>), [aggregate](<https://devfeed.tech/tags/aggregate.md>), [alert](<https://devfeed.tech/tags/alert.md>), [all](<https://devfeed.tech/tags/all.md>), [also](<https://devfeed.tech/tags/also.md>), [analytics](<https://devfeed.tech/tags/analytics.md>), [api](<https://devfeed.tech/tags/api.md>), [app](<https://devfeed.tech/tags/app.md>), [async](<https://devfeed.tech/tags/async.md>), [bus](<https://devfeed.tech/tags/bus.md>), [domain](<https://devfeed.tech/tags/domain.md>), [event](<https://devfeed.tech/tags/event.md>), [events](<https://devfeed.tech/tags/events.md>), [node](<https://devfeed.tech/tags/node.md>), [node-js](<https://devfeed.tech/tags/node-js.md>), [production](<https://devfeed.tech/tags/production.md>), [tutorial](<https://devfeed.tech/tags/tutorial.md>)

### AI overview

Domain events move independent side effects out of a core operation, but an in-process Node.js event bus does not make delivery durable or isolate runtime latency. Use an async bus that awaits handlers and logs rejected results from `Promise.allSettled`; otherwise failures can disappear, and plain `EventEmitter` does not await async handlers. Buffer events during a database transaction and emit only after commit to avoid announcing rolled-back changes. To survive a crash after commit, save an outbox row or job in the same transaction and relay it separately; make handlers safe to rerun using an event ID. Awaiting handlers adds their latency to the request, while fire-and-forget widens the crash window and may not finish on serverless. Events fit when one business fact fans out to growing independent side effects; keep essential work such as seat reservation in the transaction. The excerpt refers to an outbox follow-up for further details.

### Source excerpt

Every tutorial stops at the typed event bus. Here are the 3 production failures it skips, and the fix for each in Node.js.

## Your LLM Judge Is Untested Code

DevFeed: [Your LLM Judge Is Untested Code](<https://devfeed.tech/articles/your-llm-judge-is-untested-code-90641.md>)

Original publisher: [Read original article](<https://thetshaped.dev/p/your-llm-judge-is-untested-code-five-rules-done-right>)

Author: The T-Shaped Dev

Published: 2026-10-20T12:15:40Z

Content type: tutorial

Language: en

Sources: [The T-Shaped Dev](<https://devfeed.tech/sources/the-t-shaped-dev.md>)

Topics: [Test automation](<https://devfeed.tech/topics/test-automation.md>), [llm-reasoning](<https://devfeed.tech/topics/llm-reasoning.md>), [CI/CD](<https://devfeed.tech/topics/cicd.md>), [LLM observability](<https://devfeed.tech/topics/llm-observability.md>)

Tags: [api](<https://devfeed.tech/tags/api.md>), [eval](<https://devfeed.tech/tags/eval.md>), [evals](<https://devfeed.tech/tags/evals.md>), [llm](<https://devfeed.tech/tags/llm.md>), [tests](<https://devfeed.tech/tags/tests.md>)

### AI overview

This tutorial presents five practices for making LLM judges in agent evaluation suites more dependable: use narrow, checkable criteria; reserve model judgments for meaning rather than exact checks code can perform; aggregate binary verdicts; calibrate the judge against hand-labeled replies; and use a different model to grade the agent. It recommends running calibration in CI before scenario evaluations.

### Source excerpt

Your agent has tests. The thing grading them doesn't. Five rules for a judge you can trust.

## The Conscious Node: A Book of 157 Tips for Production Node.js Backends

DevFeed: [The Conscious Node: A Book of 157 Tips for Production Node.js Backends](<https://devfeed.tech/articles/i-wrote-a-book-on-node-js-and-never-told-you-86334.md>)

Original publisher: [Read original article](<https://thetshaped.dev/p/i-wrote-a-book-on-nodejs-and-never>)

Author: Petar Ivanov

Published: 2026-10-13T16:29:12Z

Content type: release

Language: en

Sources: [The T-Shaped Dev](<https://devfeed.tech/sources/the-t-shaped-dev.md>)

Topics: [Node.js](<https://devfeed.tech/topics/node-js.md>), [backend frameworks](<https://devfeed.tech/topics/backend-frameworks.md>), [REST API](<https://devfeed.tech/topics/rest-api.md>)

Tags: [api-design](<https://devfeed.tech/tags/api-design.md>), [backend](<https://devfeed.tech/tags/backend.md>), [express](<https://devfeed.tech/tags/express.md>), [idempotency](<https://devfeed.tech/tags/idempotency.md>), [node-js](<https://devfeed.tech/tags/node-js.md>)

### AI overview

The Conscious Node is a book about building production-ready Node.js backends, organized into 157 tips across 10 chapters. Its subjects include architecture, code design, error handling, security, testing, APIs, databases, performance, and tooling. One example explains how an Idempotency-Key header can prevent a retried order request from creating a duplicate.

### Source excerpt

The Conscious Node: 157 tips for Node.js backends that hold up in production. Not sure yet? The first three tips are free.

## Your AI Agent Has No Stack Trace. Here's How to Debug It Anyway.

DevFeed: [Your AI Agent Has No Stack Trace. Here's How to Debug It Anyway.](<https://devfeed.tech/articles/your-ai-agent-has-no-stack-trace-here-s-how-to-debug-it-anyway-65860.md>)

Original publisher: [Read original article](<https://thetshaped.dev/p/debugging-testing-ai-agents-llm-apps-4-decisions-observability-traces-correlation-ids-checklist>)

Author: The T-Shaped Dev

Published: 2026-10-06T12:15:40Z

Content type: tutorial

Language: en

Sources: [The T-Shaped Dev](<https://devfeed.tech/sources/the-t-shaped-dev.md>)

Topics: [ai observability](<https://devfeed.tech/topics/ai-observability.md>), [developer-productivity](<https://devfeed.tech/topics/developer-productivity.md>), [rate-limiting](<https://devfeed.tech/topics/rate-limiting.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [analysis](<https://devfeed.tech/tags/analysis.md>), [api](<https://devfeed.tech/tags/api.md>), [debugging](<https://devfeed.tech/tags/debugging.md>), [development](<https://devfeed.tech/tags/development.md>), [llm](<https://devfeed.tech/tags/llm.md>), [traces](<https://devfeed.tech/tags/traces.md>)

### AI overview

The article gives a four-part observability checklist for debugging production LLM applications: capture request traces, carry a correlation ID across logs and services, monitor five metrics by model and prompt version, and log structured decisions such as validation failures and fallbacks. It also discusses content capture, sampling, privacy, token-cost calculations, and cache-rate differences between providers.

### Source excerpt

How to make LLM apps debuggable: traces, correlation IDs, and the five numbers on the dashboard.

## Saved the Order, Lost the Event: Why You Need the Outbox Pattern

DevFeed: [Saved the Order, Lost the Event: Why You Need the Outbox Pattern](<https://devfeed.tech/articles/saved-the-order-lost-the-event-why-you-need-the-outbox-pattern-64459.md>)

Original publisher: [Read original article](<https://thetshaped.dev/p/outbox-pattern-why-you-need-typescript-postgresql-nodejs>)

Author: The T-Shaped Dev

Published: 2026-10-03T12:21:10Z

Content type: tutorial

Language: en

Sources: [The T-Shaped Dev](<https://devfeed.tech/sources/the-t-shaped-dev.md>)

Topics: [Publish-subscribe pattern](<https://devfeed.tech/topics/pubsub.md>)

### AI overview

This tutorial explains how to prevent lost or phantom messages when a service writes to Postgres and publishes to a message broker. It shows how an outbox table and relay keep event creation atomic with the business change, while an inbox table deduplicates deliveries in the consumer. It also covers relay locking, broker confirmations, ordering, poison messages, retention, and monitoring.

### Source excerpt

Save-then-publish drops messages without a single error in your logs. Here's the outbox + inbox pair that stops it, in plain Postgres and Node.js.

## 6 MCP Servers That Make My Coding Agent Actually Useful in 2026

DevFeed: [6 MCP Servers That Make My Coding Agent Actually Useful in 2026](<https://devfeed.tech/articles/6-mcp-servers-that-make-my-coding-agent-actually-useful-in-2026-62098.md>)

Original publisher: [Read original article](<https://thetshaped.dev/p/6-mcp-servers-that-make-my-coding-agent-2026>)

Author: The T-Shaped Dev

Published: 2026-09-29T13:57:20Z

Content type: article

Language: en

Sources: [The T-Shaped Dev](<https://devfeed.tech/sources/the-t-shaped-dev.md>)

Topics: [MSP MCP](<https://devfeed.tech/topics/msp-mcp.md>), [cursor](<https://devfeed.tech/topics/cursor.md>), [GitHub Copilot](<https://devfeed.tech/topics/github-copilot.md>), [DuckDB](<https://devfeed.tech/topics/duckdb.md>)

Tags: [cli](<https://devfeed.tech/tags/cli.md>), [coding-agent](<https://devfeed.tech/tags/coding-agent.md>), [database](<https://devfeed.tech/tags/database.md>), [github](<https://devfeed.tech/tags/github.md>), [mcp](<https://devfeed.tech/tags/mcp.md>), [mcp-servers](<https://devfeed.tech/tags/mcp-servers.md>)

### AI overview

The author describes MCP servers used with a coding agent to fill gaps in current library documentation, repository history, database schemas, and browser access. The examples include retrieving GitHub review and CI context and inspecting a staging database with read-only credentials.

### Source excerpt

The MCP servers I actually reach for when building features and shipping production apps.

## 5 Resilience Patterns for AI Agents

DevFeed: [5 Resilience Patterns for AI Agents](<https://devfeed.tech/articles/5-resilience-patterns-for-ai-agents-59173.md>)

Original publisher: [Read original article](<https://thetshaped.dev/p/5-resilience-patterns-for-ai-agents-retries-checkpoints-idempotency>)

Author: The T-Shaped Dev

Published: 2026-09-24T10:19:30Z

Content type: tutorial

Language: en

Sources: [The T-Shaped Dev](<https://devfeed.tech/sources/the-t-shaped-dev.md>)

Topics: [Resilience](<https://devfeed.tech/topics/resilience.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [agentic workflows](<https://devfeed.tech/topics/agentic-workflows.md>), [retry](<https://devfeed.tech/topics/retry.md>), [Stateful](<https://devfeed.tech/topics/stateful.md>), [Claude](<https://devfeed.tech/topics/claude.md>), [anthropic](<https://devfeed.tech/topics/anthropic.md>)

Tags: [agentic-workflows](<https://devfeed.tech/tags/agentic-workflows.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [api](<https://devfeed.tech/tags/api.md>), [error](<https://devfeed.tech/tags/error.md>), [errors](<https://devfeed.tech/tags/errors.md>), [openai](<https://devfeed.tech/tags/openai.md>), [resilience](<https://devfeed.tech/tags/resilience.md>), [retries](<https://devfeed.tech/tags/retries.md>), [stateful](<https://devfeed.tech/tags/stateful.md>)

### AI overview

This practical article presents five resilience adjustments for production AI agents, including scoped step retries, idempotent tools, checkpoints, and run budgets. It explains why long, stateful, side-effecting agent runs require different failure-handling strategies than ordinary requests.

### Source excerpt

Checkpoints, scoped retries, idempotent tools, and run budgets: how production agents survive the failure math of 20-step loops.

## 7 Identity Patterns for AI Agents in Production

DevFeed: [7 Identity Patterns for AI Agents in Production](<https://devfeed.tech/articles/7-identity-patterns-for-ai-agents-in-production-55115.md>)

Original publisher: [Read original article](<https://thetshaped.dev/p/7-identity-patterns-for-ai-agents-production-systems>)

Author: The T-Shaped Dev

Published: 2026-09-18T13:57:47Z

Content type: article

Language: en

Sources: [The T-Shaped Dev](<https://devfeed.tech/sources/the-t-shaped-dev.md>)

Topics: [AI Agent](<https://devfeed.tech/topics/ai-agent.md>), [Security](<https://devfeed.tech/topics/security.md>), [OAuth](<https://devfeed.tech/topics/oauth.md>), [Entra ID](<https://devfeed.tech/topics/entra-id.md>), [SPIFFE](<https://devfeed.tech/topics/spiffe.md>), [Authorization](<https://devfeed.tech/topics/authorization.md>), [standard](<https://devfeed.tech/topics/standard.md>)

Tags: [agent-identity](<https://devfeed.tech/tags/agent-identity.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [backend](<https://devfeed.tech/tags/backend.md>), [credentials](<https://devfeed.tech/tags/credentials.md>), [entra-id](<https://devfeed.tech/tags/entra-id.md>), [identity](<https://devfeed.tech/tags/identity.md>), [oauth](<https://devfeed.tech/tags/oauth.md>), [security](<https://devfeed.tech/tags/security.md>), [spiffe](<https://devfeed.tech/tags/spiffe.md>), [standard](<https://devfeed.tech/tags/standard.md>)

### AI overview

This article presents identity and authorization patterns for AI agents in production systems. It argues that agents should have distinct identities, credentials, permissions, and accountable human owners, and recommends delegation through OAuth token exchange so audit logs distinguish users from the agents acting on their behalf. It also discusses agent identities in Entra ID and SPIFFE-based certificate rotation.

### Source excerpt

Your Agent Isn't a User. Stop Authenticating It Like One.

## The 3 Layers of Testing AI Agents

DevFeed: [The 3 Layers of Testing AI Agents](<https://devfeed.tech/articles/the-3-layers-of-testing-ai-agents-26894.md>)

Original publisher: [Read original article](<https://thetshaped.dev/p/the-3-layers-of-testing-ai-agents-unittests-mocking-alerts-judge-evals-production-agentic-systems>)

Author: The T-Shaped Dev

Published: 2026-09-15T20:59:04Z

Content type: tutorial

Language: en

Sources: [The T-Shaped Dev](<https://devfeed.tech/sources/the-t-shaped-dev.md>)

Topics: [Software Testing](<https://devfeed.tech/topics/software-testing.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>)

Tags: [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [how-to](<https://devfeed.tech/tags/how-to.md>), [testing](<https://devfeed.tech/tags/testing.md>)

### AI overview

The article presents a three-layer approach to testing AI agents: deterministic tests for surrounding code such as tools, schemas, routing, state, and failure policies; judged scenarios for model behavior; and continuous production evaluations.

### Source excerpt

Draw the line between the model and the machine, and 80% of agent testing is testing you already know how to do. (7 min)

## When to Split a Single AI Agent into Multiple Agents

DevFeed: [When to Split a Single AI Agent into Multiple Agents](<https://devfeed.tech/articles/don-t-start-with-multi-agent-18114.md>)

Original publisher: [Read original article](<https://thetshaped.dev/p/multi-agent-is-the-new-microservices-one-agent-vs-many-agents-supervisor-pattern>)

Author: The T-Shaped Dev

Published: 2026-09-12T05:19:14Z

Content type: opinion

Language: en

Sources: [The T-Shaped Dev](<https://devfeed.tech/sources/the-t-shaped-dev.md>)

Topics: [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [software-architecture](<https://devfeed.tech/topics/software-architecture.md>), [Microservices](<https://devfeed.tech/topics/microservices.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [microservices](<https://devfeed.tech/tags/microservices.md>), [multi-agent](<https://devfeed.tech/tags/multi-agent.md>)

### AI overview

The article argues that multi-agent systems can be worthwhile when a single agent suffers from expanding tool choices, conflicting system instructions, and excessive blast radius. It recommends starting with one agent and using measurable signs to determine when to split into specialized agents managed by a supervisor.

### Source excerpt

When one agent should become many, when it shouldn't, and the supervisor pattern that makes the split work. (8 min)

## 7 Rules for Designing Tool Definitions That AI Agents Can Reliably Use

DevFeed: [7 Rules for Designing Tool Definitions That AI Agents Can Reliably Use](<https://devfeed.tech/articles/your-agent-isn-t-dumb-your-tools-are-18104.md>)

Original publisher: [Read original article](<https://thetshaped.dev/p/7-rules-tool-definitions-ai-agents-tool-design-software-design-architecture>)

Author: The T-Shaped Dev

Published: 2026-09-08T12:19:46Z

Content type: tutorial

Language: en

Sources: [The T-Shaped Dev](<https://devfeed.tech/sources/the-t-shaped-dev.md>)

Topics: [Code](<https://devfeed.tech/topics/code.md>), [Tool](<https://devfeed.tech/topics/tool.md>), [API](<https://devfeed.tech/topics/api.md>), [Vibe coding](<https://devfeed.tech/topics/vibe-coding.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [code](<https://devfeed.tech/tags/code.md>), [coding](<https://devfeed.tech/tags/coding.md>), [coding-agents](<https://devfeed.tech/tags/coding-agents.md>), [developer](<https://devfeed.tech/tags/developer.md>), [tool](<https://devfeed.tech/tags/tool.md>), [tools](<https://devfeed.tech/tags/tools.md>)

### AI overview

A tutorial on designing tool definitions for AI agents. It explains that models rely on a tool's name, description, and schema, so ambiguous contracts can cause incorrect behavior. The article recommends focused tools, descriptions that guide selection and state when not to use a tool, and structured decisions followed by deterministic code.

### Source excerpt

7 rules for tool definitions a model can't misread, starting with the sentence that fixes the most bugs. (8 min)

## Classify Domain and System Errors to Set HTTP Status, Logging, and Paging

DevFeed: [Classify Domain and System Errors to Set HTTP Status, Logging, and Paging](<https://devfeed.tech/articles/stop-throwing-new-error-it-s-lying-to-your-logs-18109.md>)

Original publisher: [Read original article](<https://thetshaped.dev/p/domain-errors-are-features-system-errors-are-bugs-error-handling>)

Author: The T-Shaped Dev

Published: 2026-08-29T04:19:38Z

Content type: tutorial

Language: en

Sources: [The T-Shaped Dev](<https://devfeed.tech/sources/the-t-shaped-dev.md>)

Topics: [Code](<https://devfeed.tech/topics/code.md>), [HTTP](<https://devfeed.tech/topics/http.md>), [API](<https://devfeed.tech/topics/api.md>)

Tags: [api](<https://devfeed.tech/tags/api.md>), [code](<https://devfeed.tech/tags/code.md>), [errors](<https://devfeed.tech/tags/errors.md>), [graphql](<https://devfeed.tech/tags/graphql.md>), [http](<https://devfeed.tech/tags/http.md>), [logs](<https://devfeed.tech/tags/logs.md>), [middleware](<https://devfeed.tech/tags/middleware.md>), [on-call](<https://devfeed.tech/tags/on-call.md>)

### AI overview

This tutorial distinguishes expected domain errors, such as invalid input or forbidden actions, from unexpected system errors, such as database failures, upstream timeouts, and memory exhaustion. It proposes encoding the distinction in error types so middleware can determine the HTTP status, user message, log level, and paging behavior.

### Source excerpt

One distinction that sets your HTTP status, your log level, and who gets paged. (6 min)

## How React Rendering Works and Why Memoization Is Often Not the First Fix

DevFeed: [How React Rendering Works and Why Memoization Is Often Not the First Fix](<https://devfeed.tech/articles/your-usememo-probably-isn-t-doing-anything-18116.md>)

Original publisher: [Read original article](<https://thetshaped.dev/p/react-rendering-demystified-your-usememo-probably-isnt-doing-anything>)

Author: The T-Shaped Dev

Published: 2026-07-28T06:10:09Z

Content type: tutorial

Language: en

Sources: [The T-Shaped Dev](<https://devfeed.tech/sources/the-t-shaped-dev.md>)

Topics: [React](<https://devfeed.tech/topics/react.md>), [User Interfaces](<https://devfeed.tech/topics/user-interfaces.md>), [Code](<https://devfeed.tech/topics/code.md>), [JavaScript](<https://devfeed.tech/topics/javascript.md>), [software-development](<https://devfeed.tech/topics/software-development.md>)

Tags: [code](<https://devfeed.tech/tags/code.md>), [component](<https://devfeed.tech/tags/component.md>), [context](<https://devfeed.tech/tags/context.md>), [javascript](<https://devfeed.tech/tags/javascript.md>), [react](<https://devfeed.tech/tags/react.md>), [render](<https://devfeed.tech/tags/render.md>), [state](<https://devfeed.tech/tags/state.md>), [ui](<https://devfeed.tech/tags/ui.md>)

### AI overview

This tutorial explains that a React render is a component function call, while DOM updates happen only during the commit phase. It identifies state changes, context changes, and new child elements from a parent as render triggers, and argues that moving state or restructuring children often helps before memoization. It also notes that StrictMode intentionally renders twice in development and that the compiler cannot move component state.

### Source excerpt

One rule explains every React re-render. Get it right and memoization becomes the last fix, not the first.

## 7 Things That Break LLM Apps in Production

DevFeed: [7 Things That Break LLM Apps in Production](<https://devfeed.tech/articles/7-things-that-break-llm-apps-in-production-18105.md>)

Original publisher: [Read original article](<https://thetshaped.dev/p/7-things-that-break-llm-apps-in-production-ai>)

Author: The T-Shaped Dev

Published: 2026-07-08T10:20:03Z

Content type: tutorial

Language: en

Sources: [The T-Shaped Dev](<https://devfeed.tech/sources/the-t-shaped-dev.md>)

Topics: [Large Language Model](<https://devfeed.tech/topics/llm.md>), [software-development](<https://devfeed.tech/topics/software-development.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [ai](<https://devfeed.tech/tags/ai.md>), [llm](<https://devfeed.tech/tags/llm.md>), [production](<https://devfeed.tech/tags/production.md>)

### AI overview

A practical guide to seven production failure modes in LLM applications, using an email-drafting agent outage to show why network calls need timeouts, cancellation, bounded retries, and resilient workflow design.

### Source excerpt

The demo works on your machine. Here's what takes it down when real users show up.

## 3 Resilience Patterns That Keep a Backend Service Up When Its Dependencies Aren't

DevFeed: [3 Resilience Patterns That Keep a Backend Service Up When Its Dependencies Aren't](<https://devfeed.tech/articles/3-resilience-patterns-that-keep-a-backend-service-up-when-its-dependencies-aren-t-18117.md>)

Original publisher: [Read original article](<https://thetshaped.dev/p/resilience-patterns-keep-backend-nodejs-service-up-production-timeout-retry-backoff-jitter-circuit-breaker>)

Author: The T-Shaped Dev

Published: 2026-06-27T10:19:01Z

Content type: tutorial

Language: en

Sources: [The T-Shaped Dev](<https://devfeed.tech/sources/the-t-shaped-dev.md>)

Topics: [Resilience](<https://devfeed.tech/topics/resilience.md>), [Back end](<https://devfeed.tech/topics/backend.md>), [API](<https://devfeed.tech/topics/api.md>)

Tags: [backend](<https://devfeed.tech/tags/backend.md>), [circuit](<https://devfeed.tech/tags/circuit.md>), [dependencies](<https://devfeed.tech/tags/dependencies.md>), [patterns](<https://devfeed.tech/tags/patterns.md>), [resilience](<https://devfeed.tech/tags/resilience.md>), [retry](<https://devfeed.tech/tags/retry.md>), [timeout](<https://devfeed.tech/tags/timeout.md>)

### AI overview

A tutorial on using timeouts, retries with backoff, and circuit breakers to limit cascading failures in backend services when dependencies become slow or unreliable. It explains how unbounded waiting can exhaust server resources and describes how to layer these resilience patterns.

### Source excerpt

Timeout, retry with backoff, and circuit breaker -- what each one actually does, and the order to stack them.

## A Practical CQRS Split for Read and Write Paths in a Node.js Service

DevFeed: [A Practical CQRS Split for Read and Write Paths in a Node.js Service](<https://devfeed.tech/articles/cqrs-without-the-astronaut-architecture-18106.md>)

Original publisher: [Read original article](<https://thetshaped.dev/p/cqrs-without-the-astronaut-architecture-split-read-write-paths-typescript-nodejs-backend-service>)

Author: The T-Shaped Dev

Published: 2026-06-21T12:19:52Z

Content type: tutorial

Language: en

Sources: [The T-Shaped Dev](<https://devfeed.tech/sources/the-t-shaped-dev.md>)

Topics: [Back end](<https://devfeed.tech/topics/backend.md>), [Code](<https://devfeed.tech/topics/code.md>)

Tags: [architecture](<https://devfeed.tech/tags/architecture.md>), [business-logic](<https://devfeed.tech/tags/business-logic.md>)

### AI overview

This tutorial presents a lightweight CQRS approach for a Node.js service: separate business-heavy write operations from read operations that serve screens, without requiring event sourcing or a second database.

### Source excerpt

You don't need event sourcing or a second database. Just split reads from writes, and finally fix the 20-method service nobody wants to open.

## When React compound components are a poor fit for data-driven lists

DevFeed: [When React compound components are a poor fit for data-driven lists](<https://devfeed.tech/articles/you-re-using-react-compound-components-wrong-18121.md>)

Original publisher: [Read original article](<https://thetshaped.dev/p/you-are-using-react-compound-components-wrong-type-safe-typescript>)

Author: The T-Shaped Dev

Published: 2026-06-16T10:19:04Z

Content type: tutorial

Language: en

Sources: [The T-Shaped Dev](<https://devfeed.tech/sources/the-t-shaped-dev.md>)

Topics: [React](<https://devfeed.tech/topics/react.md>), [Tutorial](<https://devfeed.tech/topics/tutorial.md>)

Tags: [components](<https://devfeed.tech/tags/components.md>), [react](<https://devfeed.tech/tags/react.md>), [tutorial](<https://devfeed.tech/tags/tutorial.md>), [typescript](<https://devfeed.tech/tags/typescript.md>)

### AI overview

The article argues that React compound components are often misapplied to data-driven lists. It explains that when items come from an array, a plain props-based API better supports operations such as sorting, filtering, virtualization, and keyboard navigation, while avoiding unnecessary TypeScript complexity.

### Source excerpt

The most-copied example in every tutorial is the one place you shouldn't use them.

## How to Stop Charging Customers Twice: A Practical Idempotency Guide

DevFeed: [How to Stop Charging Customers Twice: A Practical Idempotency Guide](<https://devfeed.tech/articles/how-to-stop-charging-customers-twice-a-practical-idempotency-guide-18113.md>)

Original publisher: [Read original article](<https://thetshaped.dev/p/how-to-stop-charging-customers-twice-practical-idempotency-guide-http-message-queues>)

Author: The T-Shaped Dev

Published: 2026-06-09T03:20:24Z

Content type: tutorial

Language: en

Sources: [The T-Shaped Dev](<https://devfeed.tech/sources/the-t-shaped-dev.md>)

Topics: [HTTP](<https://devfeed.tech/topics/http.md>), [systems](<https://devfeed.tech/topics/systems.md>)

Tags: [guide](<https://devfeed.tech/tags/guide.md>), [how-to](<https://devfeed.tech/tags/how-to.md>), [http](<https://devfeed.tech/tags/http.md>)

### AI overview

A practical guide to idempotency explains how to prevent duplicate charges when payment requests are retried. It covers HTTP requests, message queues, and the distinction between idempotent and non-idempotent operations.

### Source excerpt

Idempotency in plain English: HTTP, message queues, and the one detail that decides whether it's real.

## My Frontend Stack In 2026

DevFeed: [My Frontend Stack In 2026](<https://devfeed.tech/articles/my-frontend-stack-in-2026-18115.md>)

Original publisher: [Read original article](<https://thetshaped.dev/p/my-frontend-stack-in-2026-react-nextjs-pnpm-vite-ts-tailwind-storybook-tanstack-zustand-zod-oxlint-oxfmt-msw-vitest-playright-sentry>)

Author: The T-Shaped Dev

Published: 2026-05-23T02:19:53Z

Content type: opinion

Language: en

Sources: [The T-Shaped Dev](<https://devfeed.tech/sources/the-t-shaped-dev.md>)

Topics: [Front end](<https://devfeed.tech/topics/frontend.md>), [Next.js](<https://devfeed.tech/topics/next-js.md>), [Vite](<https://devfeed.tech/topics/vite.md>), [React](<https://devfeed.tech/topics/react.md>), [Tailwind CSS](<https://devfeed.tech/topics/tailwind.md>), [Shadcn UI](<https://devfeed.tech/topics/shadcn-ui.md>), [TypeScript](<https://devfeed.tech/topics/typescript.md>), [Single-page application (SPA)](<https://devfeed.tech/topics/spa.md>), [Server-side rendering](<https://devfeed.tech/topics/server-side-rendering.md>)

Tags: [2026](<https://devfeed.tech/tags/2026.md>), [frontend](<https://devfeed.tech/tags/frontend.md>), [next-js](<https://devfeed.tech/tags/next-js.md>), [react](<https://devfeed.tech/tags/react.md>), [ssr](<https://devfeed.tech/tags/ssr.md>), [tailwind](<https://devfeed.tech/tags/tailwind.md>), [tailwind-css](<https://devfeed.tech/tags/tailwind-css.md>), [typescript](<https://devfeed.tech/tags/typescript.md>), [vite](<https://devfeed.tech/tags/vite.md>)

### AI overview

An opinionated overview of a 2026 frontend stack for production applications. The author describes choosing Next.js for content-heavy sites, Vite with React and TypeScript for internal tools and single-page applications, pnpm for package management, and Tailwind CSS with shadcn/ui for styling and components.

### Source excerpt

The tools I actually reach for when building production apps, from framework to the moment a user hits a bug.

[Next page](<https://devfeed.tech/sources/the-t-shaped-dev.md?cursor=WyIyMDI2LTA1LTIzVDAyOjE5OjUzKzAwOjAwIiwgIjJkZGEzZThhLTMwMjktNGVhZi05YjY0LTAxNDFjM2RjNTk3MSJd>)