AI Coding + Open Source
Posts matching all selected tags.
Category
Tags
7 more tagsFewer tags
Newsletter
New posts land in your inbox when they publish. No spam, unsubscribe anytime.
You're in. New posts head your way.
Filters2
11 posts
Move GPT off Cursor before November 12
Codex vs Cursor after OpenAI dated the Cursor GPT pipe. Install Codex CLI, pin gpt-5.6-sol or Terra, then prove a real coding turn before November 12.
Best free coding agent in 2026, ranked and compared
The best free coding agent depends on what free actually means, BYOK forever, a rotating model window, or a paid cloud tier behind a free core, checked live.
Claude Cowork vs OpenClaw is a permission model fight
claude cowork vs openclaw is a permission default. Cowork stays in a folder and still asks to delete. OpenClaw ships host exec with ask off. Pick the disk.
claude-mem exists because Claude Code starts cold
claude-mem exists because Claude Code still starts cold. Official memory covers chat and Cowork. Decide before you treat Tuesday's launch as a CLI fix.
Install claude-mem so the next session remembers
Install claude-mem with npx, not a global npm package. Curl the per-user worker port, then start a new session that prints observation IDs without a paste.
OpenClaw vs Claude is not a coding agent bakeoff
OpenClaw vs Claude is a category error. One is a session with a permission dialog. The other is a host Gateway with ask off. Read the docs before you pick.
OpenSpec until the spec file is the source of truth
OpenSpec Claude Code init writes skills, not a spec. Propose a delta, cite that path in the PR, then archive after merge so specs/ is the source of truth.
Qwen 3.8 overthinks on the default reasoning setting
Qwen 3.8 overthinks on the default xhigh setting. A pelican SVG ate 22,276 thinking tokens and 21 minutes. Start on low or off, then raise it for hard jobs.
OpenCode vs Claude Code on the Same Tasks: Same Passes, Different Paths
OpenCode vs Claude Code on four graded tasks. Both hit 4/4. Paths diverged on tools, wall time, and reported tokens. Product lane, not a same-model lab.