Trustbut on Maven: A No-Op JUnit 6 Test Gets a PASS (2026)
Trustbut 0.7.0 on Maven: a no-op JUnit 6 test earned PASS, while adding no test earned WARN. Ten receipt cases, with the limits of what they prove.
Read moreThe Blog
In-depth guides and honest reviews covering the best AI tools available today. New articles published weekly.
All
AI Coding Agents Spring Boot 4 Migration Testing & CI AI on Real Code AI API & Errors AI Models & Basics
Spring Boot 4.2 Redis: Why database=5 Really Meant db0 (2026)
On Spring Boot 4.1.2, spring.data.redis.database=5 binds correctly, but master/replica connections are built for db0. I ran 12 cases against 4.2.0-RC2.
Read moreSpring Boot 4.2 Properties Migrator: Four Gaps That Stay Silent
Tested spring.properties.migrator.fail in 4.2.0-RC1: on-error skips renamed keys, @PropertySource is invisible, and typos stay silent.
Read moreAFTER_COMMIT Is Not Durable Delivery: A Spring Outbox Checklist
Separate transaction timing from notification durability. Design an outbox with retries, idempotency, crash recovery, and failure-path verification.
Read moreSpring Boot IDOR: Enforce Ownership on Every Order Query
Fix object-level authorization on order lookup, listing, and search. Parameterized SQL is not an ownership check; verify rejected access too.
Read moreHibernate 7 Migration: Check Schema, Query Types, and Time Values
Verify a Hibernate upgrade against existing schema and data. Separate naming, native-query mappings, temporal semantics, and performance checks.
Read moreSpring Boot 4 AGENTS.md: A Migration Verification Template
A Boot 4 instruction template that checks dependencies, property binding, test discovery, and security without treating green builds as proof.
Read moreSpring Boot Method Security: Verify @PreAuthorize Is Active
Check method-security configuration, proxy invocation, and allowed and denied identities after an upgrade. An annotation alone is not enforcement.
Read moreSpring Boot 3.5 to 4: A Runnable Verification Lab
Run a Java 17 project on Boot 3.5.7 and 4.0.0. Inspect CSRF, real HTTP error dispatch, Flyway history, and two deliberately failing checks.
Read moreReviewing Copilot Spring Boot 4 PRs: Evidence and Gaps
Two public migration PRs illustrate how to separate proposed changes, reported build results, and the evidence needed to approve a Boot 4 upgrade.
Read moreSpring Boot 4 YAML: Verify Sources, Binding, and Behavior
Audit application.yml after a Boot 4 change without treating metadata, migrator output, or an actuator response as complete proof of configuration.
Read moreGradle Tests Not Rerun: UP-TO-DATE, Cache, and Missing Inputs
Distinguish legitimate Gradle test reuse from an undeclared-input bug. Check task outcomes, force execution, and declare the state the test reads.
Read moreCodex in CI: Verify the Build and Preserve the Business Rule
Separate an agent process result from application acceptance. Preserve build exit codes, inspect fresh test reports, and resolve tests-versus-spec conflicts.
Read moreSpring Boot 4 Dependencies: Resolved Does Not Mean Configured
Diagnose a Boot 4 dependency that resolves and compiles but does not activate its feature. Check integration modules, conditions, beans, and behavior.
Read moreCodex and Testcontainers: Diagnose Docker Socket Access
Check the active Docker endpoint inside the agent environment, separate daemon availability from permissions, and verify integration tests actually execute.
Read moreFlyway Checksum Mismatch: Restore History Before Repair
Diagnose edited migrations, preserve applied SQL, and distinguish Flyway metadata repair from a real schema correction before changing a database.
Read moreSpring Boot 4 Tests Not Running: Audit Surefire Discovery
Check Surefire naming, explicit selection, Failsafe binding, and fresh reports. Avoid confusing test-source counts with executed test cases.
Read moreCodex Cannot Run Maven: Diagnose JDK, Cache, and Network Access
Compare the same Maven command inside and outside Codex, then separate JDK errors, filesystem denials, repository configuration, and network failures.
Read moreJava CI Gates: Prove ArchUnit Rules Fail on Violations
Use positive and negative fixtures to verify architecture rules, preserve formatting gates, and review agent changes to the checks themselves.
Read moreSpring Boot 4 403: Diagnose CSRF and Verify the Security Chain
Distinguish CSRF, authentication, authorization, and missing test filters. Verify a scoped security change without weakening protection for a green build.
Read moreSpring Boot 4 Migration: Choose the Contracts to Verify
Scope a Boot 4 upgrade around JSON, security, persistence and configuration contracts. Choose the right checks before following the runnable case study.
Read moreAGENTS.md for Spring Boot: Discovery, Scope, and Local Overrides
Configure Codex project instructions without confusing nested scope, local overrides, permissions, or verification of a Spring Boot change.
Read moreFilesystem MCP with Claude Code: Scope Paths and Verify Access
Configure a filesystem MCP server for a small Java workspace, verify allowed directories, and avoid assuming an unsupported flag creates read-only access.
Read moreOpenAI Codex with Spring Boot: Setup, Tasks, and Verification
Use Codex on a Spring Boot repository with explicit build commands, scoped tasks, and independent verification of tests and business behavior.
Read moreCursor Codebase Context for Spring Boot: Verify Retrieved Files
Improve Cursor context for multi-module Spring Boot projects with explicit paths, indexing checks, and verification of bean and caller assumptions.
Read moreClaude Code Memory for Spring Boot: Instructions and Verification
Keep stable Spring Boot instructions separate from local setup notes and remembered observations. Check what loaded and verify facts before reusing them.
Read moreClaude Code in GitHub Actions: Review Trusted Java Changes
Configure a Claude Code review workflow with documented action inputs, explicit trust boundaries, independent builds, and visible usage limits.
Read moreGitHub Copilot with Spring Boot: Evaluate Edits and Reviews
Use Copilot with a defined Java task, repository context, and independent tests. Compare outcomes without unsupported model rankings or time-saving claims.
Read moreAI for Java Developers: A Guide to Setup and Verification
Choose a starting point for AI-assisted Java work: agent setup, repository context, meaningful tests, security checks, and Spring Boot 4 migration.
Read moreClaude Code Subagents and Hooks: Scope Java Review Tasks
Use a small read-oriented subagent and verified hook behavior. Keep instructions, tool permissions, and operating-system boundaries distinct.
Read moreCursor Cloud Agents for Spring Boot: Environment and Acceptance
Prepare a reproducible Java environment for a remote agent and review its branch, dependencies, tests, and profile assumptions before accepting a patch.
Read moreAI Code Review for Java: Check Behavior Beyond the Diff
Review transactions, ownership, bean wiring, query behavior, and Java version assumptions using concrete failure paths and executable acceptance checks.
Read moreAI-Generated JUnit Tests: Assertions That Can Detect a Regression
Review generated JUnit tests for boundary behavior, mock scope, discovery, and failure sensitivity. Includes a downloadable Java 17 example.
Read moreCursor with Spring Boot: Rules, Context, and Verification
Use Cursor on a Spring Boot repository with scoped context, project rules, authorization checks, and a build review that goes beyond compilation.
Read moreClaude Code with Spring Boot: A Verifiable Repository Workflow
Set up a Claude Code task around your Spring Boot build, project instructions, scoped changes, and checks that establish what actually ran.
Read more429001 Errors: Identify the Gateway Before Retrying
Trace a 429001 response to its provider, distinguish Tencent TokenHub codes from OpenAI errors, and preserve the evidence needed for a useful fix.
Read moreLangChain Tool Calling: A Bounded Agent with a Local Fixture
Build a small LangChain agent using a deterministic tool, configure a compatible provider explicitly, and inspect tool results without inventing live data.
Read moreAI-Assisted PDF to Word: Review the Converter and Its Checks
Inspect an open-source PyQt PDF-to-Word converter, its heading heuristics, and why a zero suspicious-character score does not prove conversion fidelity.
Read moreUsing AI in a Legacy Java Codebase: Preserve Behavior First
Map callers, capture current contracts, and make small reviewed Java changes with AI when tests and business documentation are incomplete.
Read moreMCP Tutorial: Connect One Tool and Verify Its Boundaries
Understand MCP hosts, clients, servers, and tools. Start with one scoped integration and verify discovery, access, and failure behavior.
Read moreCursor Rules Tutorial: Scope Instructions and Check Activation
Write focused Cursor project rules, choose activation deliberately, and test whether the right guidance was supplied without treating it as enforcement.
Read moreAgent Loop Engineering: Stop, Verify, and Recover
Build a bounded coding-agent loop with observable state, independent verification, repeat detection, and explicit stops for problems it cannot repair.
Read moreOpenClaw vs Hermes Agent: What to Compare Before Setup
A source-based comparison of OpenClaw and Hermes Agent: channels, memory, model routing, tool permissions, data flow, and a repeatable evaluation plan.
Read moreAI Hallucinations in Code: Check APIs, Versions, and Evidence
Catch invented APIs and unsupported claims with version-specific documentation, compilation, runtime checks, and explicit uncertainty.
Read moreHarness Engineering: Make Agent Work Verifiable
Design the environment around an AI coding agent: clear task contracts, constrained tools, independent checks, and a handoff with inspectable evidence.
Read moreOpenAI Rate Limits: Plan Request and Token Capacity
Estimate request and token demand, pace shared workers, and use account limits and response headers without relying on outdated tier tables.
Read moreCursor Tutorial: Set Up a Project and Verify Your First Change
Start with a working repository, scope a small Cursor task, inspect context and rules, and verify the resulting code before accepting it.
Read moreOpenAI API 429: Diagnose Before You Retry
Separate temporary throttling from exhausted credits, spend limits, and gateway errors. Includes an offline Python classifier and a recovery checklist.
Read moreClaude Projects Tutorial: Organize Sources and Instructions
Set up a focused Claude project, separate reference material from instructions, and test whether answers preserve sources and uncertainty.
Read moreClaude vs ChatGPT: A Reproducible Evaluation Worksheet
Compare Claude and ChatGPT using identical writing, Java review, and document tasks. Record tools, sources, corrections, and time before choosing a plan.
Read moreChoosing AI Tools for a Java Development Workflow
Choose an assistant around a concrete Java task: code review, a bounded repository change, or documentation. Includes a reusable evaluation worksheet.
Read morePrompt Engineering: Define Inputs, Constraints, and Checks
Write prompts that preserve facts and expose missing context. Use examples, output contracts, and verification instead of unsupported performance promises.
Read moreChatGPT for Beginners: Ask, Check, and Refine Your First Task
Start with a concrete ChatGPT task, provide the needed context, check factual claims, and review privacy and plan settings without relying on old model tables.
Read more