Anthropic IDE push & control

Anthropic is testing a bigger Claude Code desktop app while a separate account dispute exposed how much control the company still holds over access. (testingcatalog.com, wwwhatsnew.com) TestingCatalog reported on April 13 that Anthropic’s internal “Epitaxy” build adds integrated-development-environment style side panels, support for multiple repositories, and a “Coordinator Mode” that can manage several Claude Code sessions at once. The same outlet had reported on March 31 that Anthropic was already testing a new Claude Code interface under the Epitaxy name. (testingcatalog.com, testingcatalog.com) Anthropic’s own product page says Claude Code can read a codebase, edit files, run tests, and produce committed code, and a December 2025 Anthropic post said the desktop app could already run multiple local and remote sessions in parallel. Epitaxy points to Anthropic pulling more of that workflow into one window instead of a terminal-style tool. (anthropic.com, anthropic.com) A coding agent is software that can inspect a repository, change code, and execute commands on a developer’s behalf; an integrated development environment is the dashboard developers use to keep files, logs, and tools in one place. Anthropic has been moving Claude Code toward that fuller workspace model, and TestingCatalog previously reported that Anthropic was also developing a web version with GitHub connections and isolated environments. (anthropic.com, testingcatalog.com) The second April 13 report came from Spanish tech outlet WWWhatsnew, which said Peter Steinberger, creator of the open-source agent framework OpenClaw and now an OpenAI employee, was temporarily banned from his Claude account after Anthropic had cut his subscription about a week earlier. WWWhatsnew said Steinberger described the API use as legitimate testing to make sure OpenClaw updates did not break Claude support. (wwwhatsnew.com) WWWhatsnew has separately described OpenClaw as a fast-growing open-source agent project, reporting 135,000 active instances on April 13 and 247,000 GitHub stars on April 2. Those figures could not be independently verified here, but they help explain why a restriction affecting its creator drew immediate attention inside the agent-tooling market. (wwwhatsnew.com, wwwhatsnew.com) Anthropic has been broadening Claude Code beyond a single coding assistant into a larger product line. Its website now lists Claude Code for Enterprise, and Anthropic Academy materials promote training around Claude Code and the Model Context Protocol, the company’s standard for connecting models to outside tools and data sources. (anthropic.com, anthropic.com, anthropic.com) That leaves two tracks running at once on April 13: Anthropic appears to be building a more centralized coding workspace, and reports about account-level enforcement are sharpening attention on who controls access when teams build on proprietary assistants. (testingcatalog.com, wwwhatsnew.com)