Skip to news
English

AI & MCP News — 2026-10-15

6 verified updates published on 2026-10-15. Each summary links to an official announcement or release.

← All news

Speakeasy v1.43.2 adds one MCP gateway URL across projects

For organizations using organization-wide sign-in, Speakeasy v1.43.2 provides one MCP gateway URL for authorized servers across projects. Access changes take effect on the next request, while each server retains its own sign-in.

Published · Source checked

Speakeasy · Official source ↗ # Permalink

GitHub's Slack and Teams assistant gains multi-repository conversations

GitHub's Slack and Teams assistant can now work across multiple repositories in one conversation, including issue creation, code changes and bug investigations. Shared Slack messages and threads can also provide context.

Published · Source checked

GitHub · Official source ↗ # Permalink

GitHub schedules three Copilot model deprecations for November 16

GitHub scheduled Copilot deprecation of Claude Opus 4.8, Claude Opus 5 and GPT-5.6 Terra for November 16, 2026. Claude Opus 4.8 remains available to individual subscribers on annual plans.

Published · Source checked

GitHub · Official source ↗ # Permalink

GitHub makes agentic autofix generally available

GitHub made agentic autofix generally available for default-branch alerts to Code Security and Advanced Security customers on Enterprise Cloud; it reruns CodeQL to check fixes before opening draft pull requests. Access requires enterprise policy approval, organization or repository enablement and a Copilot billing method.

Published · Source checked

GitHub · Official source ↗ # Permalink

Easy MCP AI v2.2.0 adds managed skills and WP-CLI tools

Easy MCP AI v2.2.0 adds managed skills delivered over MCP and a WP-CLI runner with plugin, theme and core tools. It also introduces an option to reject API keys supplied in URLs.

Published · Source checked

Easy MCP AI · Official source ↗ # Permalink

Codex 0.163.0 adds immediate steering and parallel read-only tools

Codex 0.163.0 lets follow-up messages steer an active response immediately and can run independent read-only tool calls in parallel. It also adds searchable preferences through the /config interface.

Published · Source checked

OpenAI · Official source ↗ # Permalink