» Tag
software-engineering
75 postsStudy Finds 38.9% of AI Coding Agent PRs Contain Security Smells
Large-scale study of AI coding agent PRs finds 38.9% contain security smells, with most leaked credentials introduced by human developers, not AI.
Armin Ronacher: AI coding advances even as teams lose shared understanding
Armin Ronacher compares AI-driven vibe coding to the Tower of Babel, showing how coding agents erode teams' shared technical understanding.
Software Factories: Loops, Harnesses, and the Cost of Going Dark
Loops, harnesses, and factories: why dark automation in software pipelines quietly builds comprehension debt behind green tests.
CommitBrief — AI code reviews, right in your terminal
A provider-agnostic, local-first CLI that reviews your staged changes, a historic range, or a whole GitHub pull request. Zero telemetry, no server. Free and open source.
commitbrief.comFavur Evals: a public benchmark for which AI model codes best
Favur Evals is a public, vendor-independent leaderboard comparing AI models on real software engineering tasks across eight measurable dimensions.
A Practical Checklist for Reviewing AI-Generated Code
Why AI-generated code fails differently than human code, and a prioritized checklist covering hallucinated APIs, missing invariants, and security defaults.
Merging Concurrent Agent Patches by Base Commit, Not Arrival Time
A deterministic merge protocol for concurrent AI coding agent patches, using base commit hashes, diff digests, and hunk ownership instead of arrival time.
Bridging eBPF Verifier Diagnostics with bpfix
eBPF verifier errors often hide the real failure point; bpfix locates where safety proofs break, boosting LLM-based repair success by 11-21 points.
AI Agent Latency: Turn Wait Time Into Deep Thinking
How engineers can turn AI coding agent wait times into deliberate deep-thinking sessions instead of burnout-inducing monitoring loops.
30 Days as the Only Human in an AI Dev Team: 212 PRs, Broken CI
A 30-day AI-agent coding experiment shows how self-reported 'green' CI and unreviewed PRs let critical failures hide in plain sight.
AI Coding Agents Are Absorbing the Leaves, Not the Whole Tree
SWE-bench and METR data track fast AI coding gains, but verification cost — not raw difficulty — still defines where agents stop.