Codara blog
Writing on AI-native engineering orgs
Manifestos, research, and tactical writing on building software in an AI-augmented engineering org. Written by the team building Codara.
Latest · Research
The Ticket Isn’t Enough: 21 Real Tasks on What AI Agents Actually Need
We ran the same model on 21 real tasks twice — bare ticket vs. the upstream decisions behind it — and blind-graded both against the merged PR. From the ticket alone, the agent shipped a problem on 20 of 21. Most failures compiled and looked done.
Comparison
Linear vs Codara: When Your Issue Tracker Talks to a Coding Agent
Linear is the best human issue tracker ever built. Once a coding agent is reading the tickets, the job changes — the tracker has to carry product context a machine can use, not just a clean UI. An honest comparison.
May 20, 2026 · 8 min read
How-to
How to Write a Spec an AI Coding Agent Won’t Misinterpret
AI agents fill gaps in a spec by guessing. This is the six-part template — problem, success criteria, acceptance tests, constraints, out of scope, prior decisions — with the weak version vs. the strong version for each.
May 20, 2026 · 7 min read
Manifesto
Jira Was Built for Humans Writing Code. That Assumption Is Now Wrong.
A Jira ticket is a pointer to context that lives in Slack, Confluence, and people’s heads. A human fills the gap by asking around. An AI agent can’t — and that breaks the tool in a way no amount of AI bolted on top can fix.
May 20, 2026 · 6 min read
Manifesto
The Product Context Chain: Why AI Coding Agents Forget Everything That Matters
An AI agent wiped a startup’s production database in 9 seconds. The model wasn’t broken — it just had no idea what “production” meant in that company’s world. That’s the missing layer in AI-assisted dev, and bigger context windows won’t fix it.
May 19, 2026 · 5 min read