This Week's AI Tools: Replit Free Mode, Warp Factories, Checksum AI, fx, Claude Academy

The shortlist

• Replit Free Mode — Replit says everyday Agent work such as chat, ideation, and routine tasks no longer consumes credits in Free Mode. Core subscribers get up to 30 hours of chat per month, and Core costs $20 per month. The catch is important: this is not a free tier, limits reset every five hours, and complex work can move into paid Power or Max modes. 1
• Warp Factories — Warp's new control plane coordinates coding agents across triage, specifications, implementation, review, and verification. The factory is defined as version-controlled code and is currently in closed beta; qualified organizations may receive $10,000 in factory usage, but standard pricing is not disclosed. This is for teams with recurring engineering volume, not most solo builders. 2
• Checksum AI — Product Hunt describes an AI-native continuous testing platform that generates, runs, and auto-heals end-to-end and API tests as Playwright code in your own repository. It also tries to distinguish real bugs from stale tests. The practical caveat is governance: an auto-healed test can hide a real regression if nobody reviews the change. 3
• fx by Vercel — Product Hunt lists fx as a tiny open-source coding agent, while the official repository describes a model-agnostic, Unix-like CLI and embeddable harness written in Zig. The project is Apache-2.0, supports local or cloud inference, skills, MCP, and subagents, and is explicitly experimental. 45
• Claude Academy — Anthropic's free learning hub brings together AI fluency, Claude Code, Claude Platform, MCP, and other product courses. Product Hunt recorded its launch on August 23. It is a useful shortcut for someone who is still guessing how to use Claude in a product, but it is vendor-specific learning rather than a new build surface. 67