--field-notes · featured
i let cursor agent run for 4 hours and it shipped a feature
1,200 lines of code, three dead-ends, one saved afternoon. but only after i rewrote the prompt. a real debrief on 4 hours of unattended cursor agent.
~/$ open agentical.blog — field notes from vibe-driven development
an honest blog for engineers shipping with cursor, copilot, and claude code. no hype. no fearmongering. real workflows.
hyperagent is not what i expected
airtable's ceo built an agent platform, the internet called it "chatgpt with integrations," and one reddit user spent $900 figuring out if they were right.
the cursor-to-claude pipeline that doubled my throughput
use cursor for the chunky edits, claude code for the long-running tasks. one pattern, three weeks of trial, two specific rules that made it stick.
vibecheck: is claude code actually faster than typing?
a week of timing every commit. the answer is: yes, but only for a specific kind of work — and the gap is smaller than the discourse implies.
your AGENTS.md is the new readme. here's how to write one.
AGENTS.md is the file every coding agent reads before doing anything. mine is 90 lines. here's the structure that's survived three months of daily use.
the checklist pattern for agentic runs
a pattern for writing prompts that survive long agentic runs. numbered steps, explicit verification, a single 'done' criterion. nothing else works as well.