SShortSingh.
TopTechnologyAI & MLStartupsProgrammingScienceBusinessFinanceCrypto & Web3World
Back to feed
Programming

Harness Engineering Proposes a Structured Framework for Human-AI Development Collaboration

0
·1 views

A concept called Harness Engineering has been outlined by developers on DEV Community as a structured system designed to coordinate reasoning, knowledge, workflows, and validation in AI-assisted software development. The framework centers on two core components: a Specification Harness that clarifies what is being built and why, and a Coding Harness that governs how it is built correctly and maintainably. It models engineering as a continuous loop where every activity — from intent to deployment — feeds back into organizational learning rather than being discarded. A key analytical tool within the framework is the KK/KU/UK/UU model, which categorizes knowledge gaps to help teams identify what they know, what they recognize they do not know, and crucially, what they do not yet realize they are missing. The framework aims to address the growing challenge of preserving engineering intent and accumulating institutional knowledge as AI systems take on larger roles in the development process.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

Log in to join the discussion and vote.

Log in

Related stories

0
ProgrammingDEV Community ·

Japanese Truck Driver Self-Teaching Code Finds Toilet Panel and JS Flag Work Identically

A Japanese truck driver teaching himself Python and web development documented a key insight after roughly 122 hours of self-study. While building a browser-based personality test called DPT, he added a JavaScript flag called isProcessing to block duplicate button clicks during answer processing. He noticed this mechanism mirrored the blinking light on his home toilet panel, which disables the flush button while the system is active and re-enables it only after the cycle completes. Both systems solve the same real-world problem: preventing repeated or accidental inputs from triggering unintended duplicate actions. The comparison highlights how everyday product design principles, such as debouncing user input, translate directly into software engineering concepts.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Laravel Package laravel-ai-tasks Adds Pre- and Post-Call Budget Guards for AI Spending

A Laravel package called laravel-ai-tasks introduces a two-stage budget enforcement system for applications making AI provider calls. Unlike pre-flight-only checks, the package also validates spending after each provider response, when actual token costs are known. This prevents scenarios where a tenant close to their budget limit triggers a call that exceeds it, since real cost is only confirmed upon the provider's reply. Budgets are configured per tenant or via a shared default, with a customizable resolver that identifies tenants from request headers, authenticated users, or config values. The package integrates into all dispatch paths, including synchronous calls, streaming, and queued jobs, recording costs even when a post-call budget breach occurs.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Developer cuts AI agent costs from $87 to single digits weekly with custom token router

A developer running an OpenClaw AI agent was spending $87 per week after every tool call — including trivial cron job checks — was routed to an expensive paid model. Inspired by seeing OmniRoute report one billion tokens routed at zero cost, the developer built a pre-call classification system that assigns each request a call type and directs it to the most appropriate model tier. Simple tasks like heartbeat checks are handled by a local 9B model, while complex reasoning and code generation are reserved for paid cloud models. The routing logic also addressed a reliability issue where free OpenRouter models returned truncated output instead of standard 429 errors, which had silently caused three cron job failures in May. With deterministic routing applied before any model call, weekly costs dropped to single digits and the cron failures were eliminated.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

AI Session Logger Chron Reaches 5,600 Installs, Adds Compliance and Triage Features

Chron, an MCP server that logs AI coding sessions locally with tamper-evident hash chains, has reached 5,600 installs within six months of launch. The tool works alongside AI coding assistants like Claude Code, Cursor, and Windsurf, recording every message, file change, and tool call to a local SQLite database. Beyond logging, Chron now offers deterministic compliance analysis mapped to frameworks including SOC 2, ISO 27001, the EU AI Act, and NIST AI RMF, generating printable HTML evidence packages for control owners. The team emphasizes that the tool covers only controls observable from AI session activity, and each report carries a disclaimer that coverage does not equal compliance. An upcoming release will introduce an Attention Score — a 0–100 deterministic rating per session based on signals such as secret exposure, sensitive code changes, and open compliance findings — to help teams prioritize which sessions warrant closer review.

0 comments Read more at DEV Community
Harness Engineering Proposes a Structured Framework for Human-AI Development Collaboration · ShortSingh