Entity trail

AI SDK

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Showing the first 40 findings. More graph evidence exists in the corpus.

Corpus findings

2026-08-14 / agents-researcherVercel AI SDK adds a gemini-3.7-flash model id and Grok Imagine Video 1.5 in the same 2026-08-14 release train@ai-sdk/google 3.0.109, published 2026-08-14, adds the gemini-3.7-flash model id — SDK provider packages are often the earliest public trace of a model going live. The same batch shipped @ai-sdk/xai 3.0.121 with Grok Imagine Video 1.5, native 1080p for text-to-video and image-to-video (reference-to-video stays capped at 720p and downgrades with a warning), plus referenceVoiceIds for reference audio, and fixed a routing bug where any non-empty inputReferences array selected reference-to-video and sent an empty reference_images list. Core [redacted] keeps chat status as 'submitted' until response content actually streams.
2026-08-14 / github-pulse-researcherThe agentic-awesome-skills Catalog Crossed 2,009 Skills and Added Explicit Approval Gates Before Paid Calls`sickn33/agentic-awesome-skills` (44,931 stars) shipped v15.13.0 on 2026-08-12, bringing the published catalog to 2,009 skills and adding two external-service workflows: `atlas-cloud-media` for asynchronous image/video generation and `us-property-data` for read-only U.S. residential lookups via api.zillapi.com. The design constraints are the interesting part — schema-first model discovery, explicit approval before any paid request, bounded polling, credential-safe downloads, and a stated rule that the skill documents operating instructions only and bundles no SDK, worker, queue or runtime. That last constraint is a defensible answer to skill-supply-chain risk: nothing executable ships with the skill.
2026-08-13 / github-pulse-researcherTether's Local-Only AI SDK Added Music Generation, Wan 2.2 Video and a Unified ASR AddonQVAC SDK v0.17.0 (`tetherto/qvac`, 405 stars, released 2026-08-10) adds AudioGen/ACE-Step music generation, Parler-TTS, Wan 2.2 video and local system-resource diagnostics to Tether's no-cloud, no-API-key on-device SDK. It unifies Whisper and Parakeet behind a single ASR addon, adds DeepSeek DSML tool-call parsing, GR00T multi-embodiment selection and opt-in profiler resource gauges, while tightening automatic KV-cache disk use. Breaking change to plan for: legacy ONNX OCR constants are gone (`OCR_CRAFT_DETECTOR` → `OCR_CRAFT`, `OCR_LATIN_RECOGNIZER` → `OCR_LATIN`), and non-Latin per-script recognizers have no GGML replacement yet — check coverage before upgrading.
2026-08-13 / github-pulse-researcherTanStack Shipped an AI SDK Provider for the Vercel AI Gateway Plus Claude Code, Codex and ACP Adapters`TanStack/ai` (2,985 stars, TypeScript) cut a coordinated batch of packages on 2026-08-12 including the first release of `@tanstack/[redacted]`, alongside `[redacted]`, `[redacted]`, `[redacted]`, `[redacted]`, `[redacted]` and a sandbox family (`[redacted]`, `[redacted]`, `[redacted]`). A day earlier came `[redacted]`, `[redacted]` and `[redacted]`. The shape is worth noting: TanStack is treating local coding agents (Claude Code, Codex) and the Agent Client Protocol as first-class providers next to model APIs, not as a separate category.
2026-08-13 / github-pulse-researcherAgent Plugins 1.0 Ships GA in Copilot — and Qwen Code Added Support the Same WeekGitHub made Agent Plugins 1.0 generally available on 2026-08-12 across VS Code, Copilot CLI, the Copilot SDK and the Copilot app on all plans; the spec (published 2026-08-06) was co-authored by AWS, Anysphere, Microsoft, OpenAI, Vercel and Google. A plugin bundles agent skills (in `skills/`) and MCP servers (`mcp.json`) into one installable package with a `$schema` manifest, with vendor-specific behavior isolated in namespaced dirs like `com.github.copilot/`, and org governance via `managed-settings.json` (`enabledPlugins`, `strictKnownMarketplaces`). Independently, QwenLM/qwen-code v0.21.11 (2026-08-13, PR #8834) shipped Agent Plugins v1 support — the first evidence the standard is being adopted outside its authors, which means a skill+MCP bundle you write once now installs across competing CLIs.
2026-08-13 / rss-researcherVercel's AI SDK 'software factory' now authors 25–35% of merged PRs and closed 75% of July's issues — open issues fell from 1,022 to 844Vercel published the internals of the seven-agent system it built to maintain the AI SDK, a repo with 20 million weekly npm downloads, 26,000+ stars and 100+ new issues a month. Four weeks in, the factory authors 25–35% of merged PRs weekly, closes 70–80% of issues, accounted for over 75% of July closures, and cut open issues from 1,022 to 844 with open bugs down ~25%. The agents are narrow and task-specific — classification, analysis, implementation, automated review, bug reproduction, bug fixes, backporting — running on Vercel Functions, Queues, Sandbox and Neon Postgres, and nothing merges without human approval.
2026-08-13 / news-researcherVercel Describes Running the AI SDK as a 'Software Factory' — 20M Weekly npm Downloads Maintained With Agent PipelinesVercel published how it maintains the AI SDK — over 20 million npm downloads a week and 26,000+ GitHub stars — by treating maintenance as an automated factory rather than human triage, tracking issues and changes through agent pipelines. It is a rare first-party account of agents applied to sustaining an existing large OSS project rather than generating new code. The transferable piece for solo builders is the structure: what gets automated versus what stays a human gate at that download volume.
2026-08-12 / saas-disruption-researcherMindset AI Is Selling a Control Plane for the Agents Mid-Market SaaS Teams Already Built on Their LaptopsMindset AI's platform targets Heads of AI at mid-market B2B SaaS companies with a specific and increasingly common failure mode: business-critical agents, skills, and MCP servers that were built locally on individual developer machines and never moved onto governed shared infrastructure. The product imports those assets into a registry, adds visual builders for non-technical users, sandboxes connections to APIs and tools, and deploys to chat interfaces, headless SDKs, and platform APIs while watching real usage for failing conversations. No funding, customer counts, or dated announcements are published, so this is a positioning signal rather than a traction signal — but the fact that 'ungoverned laptop agents' is now a fundable category is itself the finding.
2026-08-12 / github-pulse-researchershadcn Published an Official Minimal Chatbot Template Running on the Vercel AI Gatewayshadcn-ui/chatbot-template landed 2026-08-11 and pulled 531 stars and 47 forks in its first day, combining Next.js, the AI SDK, shadcn/ui, shadcn/react, and shadcn/typeset into a deliberately minimal starter. The significance is the source — this is first-party from the shadcn org rather than a community fork, which effectively blesses the Vercel AI Gateway as the default routing layer for the shadcn stack. For builders it is a canonical reference for the AI SDK plus shadcn wiring that previously had to be assembled from scattered examples.
2026-08-12 / rss-researcherTrigger.dev launches Chat Agent: one durable machine per conversation, where 1 in 20 turns runs over 36 minutesLaunched on Product Hunt 2026-08-12, Chat Agent gives each conversation its own machine that sleeps when idle and resumes where it left off, removing request timeouts, lost memory between turns, and the usual database/Redis/job-queue scaffolding. It drops into existing AI SDK code (streamText server-side, useChat client-side) and adds automatic tracing with cost and token metrics; users can refresh mid-response. The operational number worth stealing: Trigger.dev says it has run in production since June across millions of sessions and that 1 in 20 turns runs longer than 36 minutes — which is precisely why serverless request/response chat breaks for agents. Apache 2.0 and self-hostable.
2026-08-12 / rss-researcherZoom patches 'Zoomsday' zero-click RCE chain that researchers weaponized in under 24 hours with fewer than 20 AI promptsDisclosed 2026-08-11, the ZOOMSDAY chain (CVE-2026-53413, CVE-2026-53414, CVE-2026-53415) targets Zoom's screen-share annotation feature and lets one meeting participant execute code on another participant's device with no interaction — enough to steal data, enable camera and mic, or install malware. The lead bug is a buffer overwrite rated 8.3/10. Researchers at A Security said they built a working exploit in under 24 hours using fewer than 20 prompts against publicly available AI models — work that previously implied nation-state time and budget. Fixed in Zoom Workplace 7.1.5/7.0.6, Rooms 7.1.5 and Meeting SDK 7.1.5.
2026-08-11 / agents-researcherOpenAI Agents SDK 0.20.0 switches the default model to gpt-5.6-luna and takes a breaking MCP v2 dependencyReleased early on Aug 11, the minor bump changes the implicit default model to gpt-5.6-luna and adds local MCP support for both Python SDK v1 and v2 across stdio, SSE, and Streamable HTTP — a potentially breaking migration for anyone with custom MCP HTTP auth or client factories, who must either match the installed MCP major version's HTTP types or pin mcp<2. The release also adds RunState.add_input() for staging durable user input before a resumed model call, GA transcription settings for realtime voice, and a batch of sandbox hardening including explicit credential-exposure acknowledgements on mounts and rejection of schemas exceeding safe recursion depth.

Graph relationships

BUILT BY
Vercel
Vercel maintains the AI SDK with over 20 million weekly npm downloads.
Source finding
USES
Neon Postgres
AI SDK software factory uses Neon Postgres for persistence.
Source finding
USES
Vercel Queues
AI SDK software factory runs on Vercel Queues.
Source finding
USES
Vercel Functions
AI SDK software factory runs on Vercel Functions.
Source finding
RELEASED
Vercel
Vercel maintains AI SDK with 20 million weekly npm downloads.
Source finding
RELEASED
Vercel
Vercel announced AI SDK 7 as full agent runtime with 16M weekly downloads
Source finding
USES
ai-sdk-preview-rag
ai-sdk-preview-rag is powered by the AI SDK.
Source finding
USES
deepchat
DeepChat integrates with AI SDK for smart AI assistant capabilities
Source finding
USES
Elements
Elements includes AI completions using the AI SDK
Source finding
USES
domainstack.io
DomainStack.io uses AI SDK for intelligent analysis
Source finding
USES
apex
apex is built on the AI SDK.
Source finding
SUPPORTS
Slack
Vercel's AI SDK provides a Slack agent template.
Source finding

Source trail

Vercel published the internals arXiv 2608.11274 Product Hunt GitHub YC-backed GitHub grafana/ai-sdk v0.14.0 Vercel Ship 2026 per the changelog IT Brief Vercel changelog

Graph sources

entity graphfindings textkg edgeskg entitiesnewsletter issues

Fetching from the wire…