In an Oct 26 OpenAI case study, Greptile says its review agent now starts a dev server for a pull request, resizes the browser to an iPhone viewport and walks the app to catch bugs code reading misses, like a button that vanishes on mobile. It also feeds its codebase Knowledge Base to Codex over MCP, which it says cuts how often AI-written code gets reverted.
Released Oct 26, a Changes button in the session header opens diffs of every file the agent touched in the last turn or the whole session, alongside a project file tree, a Ctrl+backtick terminal and a settings page for Ollama and LM Studio models. It also fixes resuming sessions with transcripts over about 8 MB and stops change checks from spawning thousands of git processes in big untracked workspaces…
Released Oct 26, editors can now list, read and import hosted Kilo cloud sessions via the local API ( GET /kilo/cloud sessions , POST /kilo/cloud/session/import ) without ever touching the session's provider credentials. Heads up: after an MCP SDK security bump, remote MCP servers whose URL redirects to another origin no longer connect, so point Kilo at the server's final URL…
Released Oct 26, agents on the Everruns platform can now page through and search huge MCP tool catalogs within a fixed schema budget, accept MCP Tasks so long tool calls don't time out, and sync their source two way with a GitHub branch. Self hosters get everruns docker up and a Helm chart, but note that session secrets are now write only to agents and the Deno Deploy sandbox is gone…
As of Oct 26, OpenAI's newest model is rolling out to Copilot Pro, Pro+, Max, Business and Enterprise in the model picker across VS Code, Copilot CLI, the cloud agent, JetBrains, Xcode and more, pitched for terminal and well scoped coding tasks at a lower cost than GPT 6.1 Sol. GitHub says it beat GPT 5.6 Terra on terminal and C tasks in early testing, and it's billed at provider list pricing under usage based billing…
Shown on HN Oct 26, the AGPL 3.0 beta stores every conversation turn verbatim on your own PC with secrets masked, then hands any MCP connected agent a 3k token packet of the turns that answer a question about the past; the author reports 88.8% on LongMemEval. Install with pip install graph mind memory && graph mind install , or claude mcp add graph mind memory python m graph mind.mcp server for tools only; it's used daily on Windows, while macOS and Linux only pass CI so far…
Spotted something we missed? Start a thread.