Developer builds 120-line JSON ledger to coordinate Claude and Codex AI pipeline sessions

·1 views

A developer running a multi-session AI content pipeline faced a coordination challenge after three months: Claude and Codex operate in separate, non-overlapping GitHub Actions workflows, making real-time handoffs impossible. To solve this, they built a 120-line Node.js script that uses a JSON file committed directly to the repository as a persistent coordination ledger. Each task record tracks status, timestamps, and polling intervals, allowing Claude to check whether Codex has completed a review before proceeding to publish. The pull-based design avoids the need for external services, authentication tokens, or network dependencies, since the repo already serves as shared state. Over 114 article review handoffs, the system has operated without a single missed coordination.

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)

Well-designed AI harness beats stronger models in multi-agent systems

Software teams are discovering that a poorly structured workflow undermines even the most powerful large language models, wasting tokens, time, and output quality. A well-engineered harness — covering role orchestration, context management, integration workflows, and anti-drift strategies — consistently outperforms a setup that relies solely on a high-end model. The recommended approach splits responsibilities between a planner agent, which handles high-level design using a frontier model, and cheaper worker agents that execute narrowly defined tasks. This division prevents context overload and agent drift, where a single agent loses sight of overall goals while handling low-level details. In practical comparisons, the planner-frontier plus worker-economy configuration achieved equivalent functional results at dramatically lower token costs than using a frontier model for both roles.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

ES2024 Introduces Promise.withResolvers to Simplify Deferred Promise Patterns

JavaScript developers have long used a workaround of leaking resolve and reject functions out of the Promise constructor to control resolution from external event listeners or callbacks. ES2024 introduced Promise.withResolvers, a static method that returns the promise, resolve, and reject as a plain object, eliminating the need for that closure escape pattern. The new API produces identical runtime behavior to the old approach, including the same microtask timing and error propagation, but makes the intent of creating a deferred Promise explicit. Common use cases include wrapping event-based APIs, building async queues where producers and consumers are decoupled, and batching operations where multiple callers await a single shared resolution. The change reduces multi-line boilerplate to a single destructuring call, making code easier to read and reason about.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

WebAssembly Enables Private, Server-Free File Conversion Directly in the Browser

A developer exploring client-side file conversion found WebAssembly (WASM) to be a viable alternative to server-based APIs, eliminating storage costs, privacy risks, and traffic bottlenecks. Most format conversion logic is written in C/C++ or Rust, and compiling it to WASM delivers near-native processing speeds within the browser tab. However, the approach comes with practical challenges including memory limits, partial threading support, large bundle sizes, and incomplete format coverage. For common formats such as MP3, WAV, AAC, and standard image and video types, browser-side conversion works well for moderate file sizes while keeping files entirely on the user's device. Developers adopting this approach are advised to allocate extra time for memory and file-header issues, which typically surface only during testing with real-world files.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

How to Automate Ubuntu Server Maintenance Using Cron Jobs and Bash

A technical guide published on DEV Community outlines how to automate routine Ubuntu server maintenance using a custom Bash script scheduled via cron jobs. The script handles package list updates, package upgrades, removal of unused packages, APT cache cleaning, and optional Docker resource pruning. A file lock mechanism is included to prevent multiple instances of the script from running simultaneously, avoiding conflicts with the package manager. All maintenance activities are written to a dedicated log file at /var/log/system_maintenance.log for auditing purposes. The guide also covers configuring sudoers to allow passwordless execution of required commands, enabling the cron job to run without manual intervention.

0 comments Read more at DEV Community