A
Adobe Firefly Video added to the directory
Adobe’s commercially-safe video model — not Firefly image, not Premiere
The AI Changelog
Directory listings appear here the day we ingest them. Curated model and feature notes sit on top of that feed.
A
Archify listed as a Cursor architecture add-on
Archify is an agent skill that compiles typed JSON into HTML/SVG system maps for Cursor, Claude Code, Codex, and OpenCode. Directory listing, not a CanIUse matrix row.
Sourcea
ast-grep listed as structural search for agents
ast-grep is a Rust CLI for AST search, lint, and rewrite. Agents can match syntax instead of regex across a repo.
SourceC
Clipchamp added to the directory
Microsoft browser and Windows video editor — Copilot lives inside it
C
Codex added to the directory
OpenAI’s terminal coding agent
C
Context7 MCP listed as a library-docs add-on
Context7 pulls version-specific documentation into Cursor, Claude Code, Codex, and OpenCode via the ctx7 CLI (skills or MCP).
SourceD
Dreamina added to the directory
ByteDance international gen studio — not CapCut
G
Gemini Omni added to the directory
Multimodal video gen and conversational edit — not Veo 3.1
G
GitIngest listed as a GitHub-to-prompt extract
GitIngest turns a public GitHub repository into a prompt-friendly extract. Also works by swapping hub for ingest in a GitHub URL.
SourceG
Google Flow added to the directory
Google Labs AI filmmaking studio — the editor, not Veo
G
Google Flow Music added to the directory
Lyria music studio — not Flow video
G
Google Vids added to the directory
Workspace video for work — not Flow, not Veo
M
MiniMax Design added to the directory
Local multimodal agent studio — not Hailuo
G
Official GitHub MCP listed as an agent add-on
GitHub’s official MCP server exposes issues, pull requests, and repository metadata. Public image: ghcr.io/github/github-mcp-server.
SourceO
OmniHuman added to the directory
ByteDance full-body avatar model — not Seedance, not CapCut
O
OpenCode added to the directory
Open-source AI coding agent for the terminal
P
Playwright MCP listed as a browser add-on
Microsoft’s Playwright MCP server lets a coding agent drive a real browser from Cursor, Claude Code, and Codex.
SourceR
Repomix listed as a repo packer for agents
Repomix packs a git repo into one XML or Markdown file for Cursor, Claude Code, Codex, and other agents. Context packing, not editing the tree.
SourceS
Seedance 2.5 added to the directory
ByteDance Seed video model — hosted in CapCut and Dreamina
T
Tavus added to the directory
Human computing video — Phoenix rendering, Raven perception, Sparrow turn-taking
S
Vercel skills CLI listed as the skills installer
The skills CLI installs agent skills into Cursor, Claude Code, Codex, and OpenCode. It is the package manager for skills, not a skill itself.
Source✨
Gemini 3.8 Flash and Flash Cyber
Google released Gemini 3.8 Flash as its workhorse reasoning and coding model at $0.75/$3.75 per million tokens, plus Gemini 3.8 Flash Cyber for trusted defenders in the Fairwind Program. Flash is in the Gemini app, Search AI Mode, Gemini API, Antigravity, and Gemini Enterprise.
SourceC
Careforce AI added to the directory
Voice and text agents that call patients to book, confirm, and reschedule visits
D
Daybreak added to the directory
OpenAI's cybersecurity service for defensive security teams
D
DeepSeek Harness added to the directory
Open-source, plugin-first framework for building coding agents on DeepSeek models
E
Encore AI added to the directory
Voice and chat agents trained on a company's top-performing reps to grow revenue from customer calls
I
Ivo added to the directory
AI contract review platform that now covers the whole deal, not just redlining
M
Muse Code added to the directory
Meta's terminal-based coding agent for large codebases
N
Natural added to the directory
Payment infrastructure built for AI agents to hold funds, pay, and get paid
🧠
Grok 4.6 released
SpaceXAI released Grok 4.6 for long-running agents and interactive visual work. It is available in Cursor, Grok Build, the API, and partners, with pricing from $2 per million input tokens and $6 per million output tokens.
SourceC
Cloudflare Kitesurf added to the directory
A cloud browser built for AI agents, not humans
B
BrowserStack Test Companion added to the directory
Agentic AI for software testing, built into the IDE
☄️
Comet added to the directory
Perplexity's AI-native browser
✨
Genspark added to the directory
No-code Super Agent for multi-step task automation
🛸
Google Antigravity added to the directory
Google's agentic development platform for autonomous coding
🧭
Kiro added to the directory
AWS's spec-driven agentic IDE
🧠
Claude Opus 5 launched
Anthropic released Claude Opus 5 as the default on Claude Max and the strongest model on Claude Pro, priced at $5/$25 per million tokens. It is available on all Claude platforms and as claude-opus-5 on the API.
Source🎨
Midjourney V8.2 is the default
Midjourney made V8.2 the default version, with a focus on aesthetics, image quality, and Personalization, plus the Edit Model that replaces Omni Reference, Character Reference, and Retexture.
SourceO
OpenAI Presence added to the directory
OpenAI's managed platform for enterprise voice and chat AI agents
A
Algolia AI Search added to the directory
API-first semantic and keyword search for AI experiences
A
AMD ROCm added to the directory
Open software platform for AI development and GPU computing
B
Baidu Qianfan added to the directory
Enterprise platform for building AI agents and model-powered apps
B
Braintrust added to the directory
AI observability and evaluation platform for production systems
C
Cloudflare AI Gateway added to the directory
AI application observability, controls, and multi-provider routing
🤖
GPT-5.5 released in ChatGPT and Codex
OpenAI released GPT-5.5 for Plus, Pro, Business, and Enterprise in ChatGPT and Codex, aimed at agentic coding, computer use, and multi-step knowledge work. GPT-5.5 Pro followed for Pro, Business, and Enterprise; the API opened the next day.
Source🎞️
Veo 3.1 Lite model card
Google DeepMind published the Veo 3.1 Lite model card. The lighter Veo 3 series model generates high-resolution video with audio from text or image prompts and is distributed through Vertex AI and the Gemini API.
Source