Show HN: Intent Layer: A context engineering skill for AI agents

The maker of “Intent Layer” says they’ve given AI assistants a literal guidebook for your code: simple AGENTS.md files that tell bots where the real logic lives and what to avoid. Fans cheered the demo where the agent stopped burning its entire “reading budget” and beelined to the bug. But the comments quickly split into two camps. One side shouted: hallelujah, focus the bot’s mission! Agent_Builder argued most AI mistakes aren’t dumb-model problems, they’re “intent drift”—the bot helpfully wanders off. Their fix: make intent explicit at every step and treat it like a rule, not a suggestion. The other side went full receipt-checker. moonshotideas praised the polish, then accused it of being “more or less a direct copy,” demanding louder credit to the original framework and dropping a link to intent-systems. Cue the copycat vs. remix debate. The memes were chef’s kiss: “GPS for confused code bots,” “Clippy’s Lonely Planet: Your Repo,” and “AGENTS.md is the ‘Do Not Press’ sticky note for robots.” Whether you call it context engineering or just common sense, the crowd agrees on one thing: AI needs a map—just don’t forget who drew it first.