Daily Kickoff

About

Up to seven things a day for people building with AI. Fewer when there aren't seven.

A quiet day — some candidates were dropped before publishing. 6 today.

  1. MCP Goes Stateless, and Developers Ask Whether That Just Makes It an API Again

    The MCP 2026-07-28 specification drops the initialize handshake and session header, replacing them with two required headers, Mcp-Method and Mcp-Name, letting gateways route and rate-limit tool calls without parsing the JSON body. Developer reaction split, with some calling it a return to plain REST. Anthropic says MCP has passed 400 million monthly SDK downloads, a fourfold increase this year.

    InfoQ AI/ML agentic coding

    #
  2. TDD inside the agent loop - theater or actual value?

    My colleagues at Thoughtworks tend to be big fans of Test-Driven Development, and many people in the industry advocate telling LLM agents to use TDD when building software. Birgitta Böckeler was curious if this really makes a difference, so conducted a few experiments .

    Martin Fowler

    #
  3. Anthropic’s watermark survives copy-paste, but not the real dev workflow

    Anthropic will embed invisible watermarks in text from Claude models launched in the EU on or after August 2, 2026, extending to the Claude API, Claude Code and cloud partners including AWS, Google Cloud and Microsoft Foundry. The mark survives copy-paste but, per GPTZero CTO Alex Cui, not intense paraphrasing that combines word-choice and syntax changes. Anthropic has not disclosed how the watermark works.

    The New Stack enterprise & governance

    #
  4. A Zoom Screen-Sharing Bug Let Anyone Take Over Other Devices on a Call

    Researchers at A Security say they used public AI models to find and exploit a Zoom vulnerability in under 20 prompts, in the screen-sharing annotation protocol, affecting Windows, macOS, Linux, iOS and Android. It could have let anyone on a call silently take over another participant's device with no interaction needed. Zoom has issued server- and client-side fixes.

    Wired AI enterprise & governance

    #
  5. Another OpenAI executive takes off

    Brad Lightcap, OpenAI's special projects lead and former COO, is departing after eight years to start "something new," per an internal memo posted to X. He stepped down as COO in April 2026 for a special-projects role reporting to CEO Sam Altman, with CRO Denise Dresser taking over much of his former role. The exit follows AGI chief Fidji Simo's July departure and CMO Kate Rouch's in April.

    The Verge AI engineering leadership

    #
  6. Arize AX adds native support for OpenTelemetry GenAI semantic conventions

    Arize AX now natively maps OpenTelemetry's gen_ai.* semantic conventions into its OpenInference trace format at ingestion, so platforms like Microsoft Agent Framework and CrewAI Studio can send telemetry directly over OTLP without a custom conversion processor. It covers model and provider metadata, messages, token usage and tool calls, enabling evaluations, cost and debugging workflows.

    Arize Blog developer experience

    #