Every processed story in chronological order, with the newest coverage first. Filter by tag, source, or score to drill in.
The attack demonstrates that AI coding agents wired into external tools via MCP create a new remote code execution surface that existing security controls — EDR, firewalls, IAM, VPNs, and even explicit agent instructions — do not catch, and that no vendor has yet claimed ownership of the fix.
Raidho's benchmark demonstrates that separating reasoning from execution across providers — combined with VSA memory instead of RAG — can match full tool-loop output quality at ×2.6 lower cost on the same task.
The benchmark reveals that dialogue capability is a distinct dimension of coding agent performance not captured by existing autonomous-system evaluations, exposing a gap between how agents are benchmarked and how they are actually used.
The workflow shows how `codex exec`'s non-interactive mode turns a conversational AI tool into a scriptable automation primitive, enabling a concrete split between exploratory and repetitive coding work without requiring a single unified tool to do both well.
GLM-5.2's combination of a 1 million token context window, expected MIT-licensed open weights, and ~$8/month pricing places a near-frontier coding model within reach of developers who cannot afford or prefer not to use Claude or Codex pricing tiers.
The release of `pi-coding-agent` as both a CLI tool and a buildable library highlights a real, unresolved packaging challenge for globally distributed npm tools in a post-shrinkwrap world.
The post demonstrates a concrete case where an AI coding agent autonomously shipped a complete feature — database migration and all — to a production codebase, with the "proof-of-work" screenshot/live-URL mechanism replacing the traditional human review step.
The sandboxed execution environments directly address a concrete risk of agentic coding workflows — agents making unwanted or destructive changes to a developer's local machine — by isolating Copilot's tool execution both locally and in GitHub-hosted environments.
MiMo Code's parallel sampling and selection approach demonstrates a concrete, measurable tradeoff — a 10–20% SWE-Bench Pro gain at 4–5× token cost — for improving reliability in long-horizon agentic coding runs where compounding step errors and context degradation are otherwise unmitigated.
Janus removes the manual step of narrating browser and terminal activity to a coding agent by piping that context directly into Claude via a local MCP server.