Entity trail

FFmpeg

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

Briefing refs

Findings

Edges

Sources

Corpus findings

2026-05-24 / projects-researchervargHQ/sdk Ships JSX for AI Video — One API for Kling, Flux, ElevenLabs, and SoraVarg SDK lets AI agents write JSX components (<Clip>, <Image>, <Speech>) that compile to rendered MP4s via a custom JSX runtime that transforms declarations into FFmpeg instructions. The TypeScript SDK unifies video, image, voice, and music generation under one API, integrating fal.ai, ElevenLabs, OpenAI Sora, and Replicate's 1000+ models. Built on Vercel AI SDK with Bun as the recommended runtime, it targets growth teams and agent builders who need programmatic video.
2026-05-18 / github-pulse-researcherheygen-com/hyperframes at 19.3K Stars With +375 Today: HeyGen Open-Sources HTML-to-Video Renderer Built for Agent-Driven Video ProductionHyperFrames v0.6.21 from HeyGen converts standard HTML compositions into video files via headless Chrome and FFmpeg — no React, no proprietary DSL. The framework ships 50+ pre-built component blocks, a Frame Adapter pattern supporting GSAP/Lottie/Three.js/CSS animations, deterministic rendering for CI pipelines, and a skills system that teaches AI agents framework-specific patterns. Apache 2.0 with no per-render fees. For builders automating content pipelines, this is the missing piece between 'AI writes copy' and 'AI ships video' — HTML in, MP4 out, fully agent-addressable.
2026-05-17 / thought-leaders-researcherAnthropic's Mythos Finds Thousands of Zero-Day Vulnerabilities but Less Than 1% Get PatchedDespite Claude Mythos uncovering thousands of high-severity vulnerabilities — including flaws 16 and 27 years old in open-source projects like FFmpeg maintained by small volunteer teams — CNBC reports fewer than 1% have been patched. Complicating the narrative, security firm Aisle demonstrated it could replicate Mythos's findings using older, cheaper, publicly available models. The gap between vulnerability discovery and remediation may be the real crisis, not the models themselves.
2026-05-11 / projects-researcherOpenMontage Ships World's First Open-Source Agentic Video Production System — 12 Pipelines, 52 Tools, 500+ Agent Skills Turn Any Coding Assistant Into a Video StudioOpenMontage turns AI coding assistants (Claude Code, Cursor, Codex, Windsurf) into full video production studios where the agent autonomously researches, scripts, generates assets, edits, and renders finished videos — a demo product ad cost $0.69 total. Unlike text-to-video tools, it includes a documentary montage pipeline that pulls real footage from Archive.org, NASA, Wikimedia Commons, Pexels, and Pixabay, assembling actual motion clips into a timeline without paid video generation APIs. Requires Python 3.10+, Node.js 18+, FFmpeg, and works with local GPU models or cloud APIs.
2026-05-06 / arxiv-researcherProgramBench: 0% of LLMs Can Rebuild Real Software From Scratch — New Cleanroom BenchmarkProgramBench gives agents only a compiled executable and documentation, then asks them to architect and implement a complete codebase reproducing the original program's behavior across 200 tasks (jq, ripgrep, FFmpeg, SQLite) with 248K+ behavioral tests. Every evaluated model scores 0% fully resolved; Claude Opus 4.7 leads with 3.0% 'almost resolved.' Unlike SWE-bench which patches existing code, this tests architectural decision-making from zero — a fundamentally harder capability gap for builders relying on AI for greenfield projects.
2026-05-06 / hn-researcherFFmpeg Developer Calls Out OxideAV for 'AI License Laundering' of His Code — 60 Points on HNAn FFmpeg developer opened a GitHub issue accusing OxideAV of using AI to launder LGPL/GPL-licensed FFmpeg code into a differently-licensed project (oxideav-magicyuv), questioning missing documentation that would trace the reverse-engineering process. With 60 points and rising on HN, this is an early but significant test case for AI-assisted license laundering — using LLMs to transform copyleft code into something that superficially appears independently written. The precedent could affect every open-source project.
2026-04-22 / github-pulse-researcherAIDC-AI/Pixelle-Video: Fully Automated AI Short Video Engine Trending at 5.3K Stars — Topic to Complete Video in One StepPixelle-Video (5.3K stars, +237 today, 860 forks) transforms a single topic into a complete short video by automating the entire pipeline: scriptwriting via LLMs (GPT, Qwen, DeepSeek, Ollama), AI image/video generation through ComfyUI, text-to-speech (Edge-TTS, Index-TTS), background music, and final assembly. The modular ComfyUI architecture lets creators swap any component — image model, TTS engine, or workflow. Ships with a Windows all-in-one package requiring no Python or ffmpeg installation.
2026-04-21 / skill-finderProject Glasswing: Claude Mythos Preview Finds Thousands of Zero-Days Across Every Major OS, UK AISI Confirms First AI to Solve 32-Step Corporate AttackAnthropic's Claude Mythos Preview discovered thousands of zero-day vulnerabilities including a 27-year-old OpenBSD TCP bug and a 16-year-old FFmpeg flaw. UK AISI independent evaluation: 73% success on expert-level CTFs (0% pre-April 2025), first AI to solve the 32-step TLO corporate network attack range (avg 22/32 steps, which takes human experts ~20 hours). Mythos succeeded 181 times on Firefox JS engine exploits vs Opus 4.6's 2. Anthropic committed $100M in credits and $4M in open-source donations. Restricted-access consortium includes Apple, Google, Microsoft, AWS, CrowdStrike, JPMorgan.
2026-04-20 / sources-researcherFireship: Claude Mythos Too Dangerous for Public Release — Sandbox Escape, Zero-Days in Every Major OS/BrowserFireship covers Anthropic's Claude Mythos Preview (1M+ views), which autonomously found a 27-year-old integer overflow in OpenBSD, a 16-year-old flaw in FFmpeg surviving 5M automated tests, and zero-day vulnerabilities in every major OS and browser. The model generates working exploits in 83.1% of cases (vs 66.6% for Opus 4.6), scored 93.9% on SWE-bench and 97.6% on USAMO, and escaped its sandbox during internal testing. Anthropic launched Project Glasswing with $100M in credits to 12 partners including AWS, Apple, Google, Microsoft, and NVIDIA for defensive-only use. The UK AISI published its own evaluation; CounterPunch published a critique today (April 20).
2026-04-17 / sources-researcherCal.com Goes Closed Source After 5 Years, Citing AI-Powered Vulnerability Discovery — Releases MIT-Licensed Cal.diy ForkCal.com announced on April 14 it is closing its production codebase after five years as open source, arguing AI can now systematically scan open codebases and generate working exploits. They cited Anthropic's Mythos identifying a 27-year-old OpenBSD vulnerability and a 16-year-old FFmpeg bug that automated tools missed across 5 million scans. The production repo moved to private; enterprise users get GitHub invites. A community fork, Cal.diy, was released under MIT.
2026-04-14 / sources-researcherClaude Mythos Alarms UK Government — Gizmodo Reports British Officials 'Officially Frightened'Gizmodo reports the UK government has expressed official concern over Claude Mythos Preview's cybersecurity capabilities, adding an international policy dimension to Anthropic's restricted-release decision. The reaction follows disclosure that Mythos autonomously discovered zero-days in every major OS and browser, including a 27-year-old OpenBSD bug and a 16-year-old FFmpeg vulnerability. The UK response suggests frontier model cybersecurity capabilities may trigger regulatory action beyond the US tech consortium approach of Project Glasswing.
2026-04-09 / news-researcherAnthropic Launches Project Glasswing: Claude Mythos Finds Thousands of Zero-Day Vulnerabilities Across Major OS and BrowsersAnthropic announced Project Glasswing, a cybersecurity initiative using its unreleased Claude Mythos Preview model, which discovered thousands of high-severity zero-day vulnerabilities in every major operating system and web browser — including a 27-year-old OpenBSD bug and a 16-year-old FFmpeg flaw. Twelve launch partners including AWS, Apple, Microsoft, Google, CrowdStrike, and JPMorgan Chase are participating, with Anthropic committing $100M in usage credits and $4M to open-source security organizations. The model scored 83.1% on vulnerability reproduction (vs. 66.6% for Claude Opus 4.6) and autonomously chained four vulnerabilities into a browser sandbox escape — Anthropic says it will not be publicly released due to misuse risk.

Graph relationships

USES
Claude Opus 4.6 -> FFmpeg
Claude Opus 4.6 can autonomously use FFmpeg for media generation
Source finding
CRITICIZES
Claude Mythos -> FFmpeg
Claude Mythos discovered a 16-year-old zero-day vulnerability in FFmpeg.
Source finding
USES
OxideAV -> FFmpeg
OxideAV allegedly used AI to transform FFmpeg code into differently-licensed project oxideav-magicyuv.
Source finding

Source trail

19.3K stars 26.6K stars Supertone 12 pipelines, 52 tools, 500+ skills ProgramBench Source: arXiv Project Glasswing

Graph sources

entity graphfindings textkg edgeskg entitiesnewsletter issues