A developer argues that AI coding harnesses like Claude Code, Codex, Antigravity, and OpenCode can be used for deterministic data engineering by having agents author artifacts at development time rath…
A developer built an autonomous AI agent for their NOC ticket queue in a single shift, using OpenCode to handle duplicate detection, categorization, and network diagnostics. The agent, now running as …
Browser-act, a new open-source CLI tool designed specifically for AI agents, enables automated browser control with anti-bot bypass, human handoff, and parallel multi-account isolation. The tool, avai…
OpenCode V2 compaction is a durable checkpoint-and-retry mechanism, not a relevance-pruning system. It is a last-resort survival mechanism that compresses all older context into a hardcoded 4,096-toke…
Domain-specific large language models have largely failed to gain traction due to the loss of generality, but a new approach using domain-specific harnesses is emerging. The harness manages memory, re…
A developer switched from the Ponytail AI coding tool to Guardsman after a race condition in Ponytail-optimized code double-charged a customer. Ponytail reduces code volume by ~54% but lacks risk awar…
RepoMap, a new open-source tool, generates interactive architectural maps of code repositories without sending source code to an LLM, reducing token usage and enabling faster architectural reasoning f…
A developer has compiled a list of recommended Effect repositories for use in opencode references, including canonical sources like Effect-TS/effect and community projects such as lucas-barake/effect-…
A developer built a fully automated coding loop in OpenCode, achieving surprisingly good results with the concept of Loop Engineering. The project demonstrates an automated workflow for coding tasks u…
SpaceXAI has open-sourced Grok Build, the terminal-based AI coding agent behind its grok CLI, under the Apache 2.0 license on GitHub. The release includes the agent harness, TUI, CLI shell, and develo…
Heimdall MCP version 1.5.0 introduces resource locks to prevent concurrent agent sessions from silently overwriting each other's changes. The feature locks tool calls by resource argument or literal k…
A developer's detailed trace of GitHits' MCP server revealed that a simple changelog query took 9.5 minutes and 91 tool calls, compared to about one minute via plain web fetching, prompting the small …
Zerostack, a minimal coding agent written in Rust and inspired by pi and opencode, launches with a 26MB binary and ~30k lines of code, consuming roughly 16MB of RAM on average versus ~300MB for JS-bas…
Thinking Machines Lab released Inkling, a 975-billion-parameter Mixture-of-Experts model that the company explicitly says is not the strongest available, positioning it as a platform for fine-tuning r…
Databricks announced Thinking Machines Lab's open-weights model Inkling is now available on the Databricks platform through Unity AI Gateway, enabling enterprise customers to build agents and applicat…
A developer's analysis of the AI-agent ecosystem reveals that model launches are the strongest acquisition event for agent tools, with openai/codex doubling claude-code's star pace within hours of GPT…
LLM-Wiki v0.16.0 introduces Query Lite, a compact read-only query protocol that reduces Claude Code wiki query instructions by 72.54%, alongside token benchmarks and reproducible benchmark lanes for C…
Dax Raad, co-founder of OpenCode, revealed on the Syntax podcast that his team's AI token usage increased 5x after gaining early access to an unreleased AI model, and when access ended, team members m…
New testing by AI consultancy Systima found that agent harness configuration can significantly inflate token overhead and enterprise AI deployment costs, with research showing that changing the harnes…
Harness Engineering built a fully automated coding loop inside OpenCode using DeepSeek-V4-Pro and DeepSeek-V4-Flash, enabling low-cost agentic code iteration. The loop, inspired by Andrew Ng's breakdo…