What main hasn't merged yet.

A twice-weekly brief on the tools, AI capabilities and shifts in the economics of software that reach developers before they become headlines. By SiyadDev.

Read the latest issue RSS feed

Disruption Watch

Where each shift stands, updated 26 October 2026

Each shift climbs from Watching to Repricing the work as the evidence builds. See the evidence for each one.

Latest issue

Issue 5, 26 October 2026

Agent Skills go neutral, Claude Code closes fail-open gaps, and MCP retries get a stop sign

Anthropic handed the Agent Skills standard to the Linux Foundation's Agentic AI Foundation, and Claude Code 2.1.308 to 2.1.310 fixed several places where safety checks failed open. One new entry on the Watch board; nothing else moved.

Read issue 5

Earlier issues

22 October 20264. Claude Code logs every blocked tool call, MCP retries get selective, and fork subagents get a fix
19 October 20263. Fork subagents are on by default, open weights split in two, and Claude Design gets an end date
15 October 20262. Hooks can now steer subagents, MCP permission rules got stricter, and Git is being rebuilt for agents
12 October 20261. Decision models arrive, guard hooks learn to fail closed, and Haiku 5.5's price has a catch

What main hasn't merged yet.

Disruption Watch

Watching

Accelerating

Hitting the market

Repricing the work

Latest issue

Agent Skills go neutral, Claude Code closes fail-open gaps, and MCP retries get a stop sign

Earlier issues