One command boots a jailbroken iPhone on your Mac — no hardware required
vphone-cli spins up a real iOS device as a VM using Apple's own Virtualization.framework, no physical iPhone needed and zero risk to your actual device. A single command handles the whole pipeline: download, patch, DFU restore, custom firmware install, first boot.
What makes it interesting is the choice: five patch variants ranging from stock security all the way to a full jailbreak, so you pick exactly how deep the bypass goes. Pick the top variant and Sileo plus TrollStore install themselves on first boot — then SSH straight into the shell.
It's a Swift CLI, Apple Silicon and macOS 15+ only. Clone the repo, run the setup and build scripts, then
https://github.com/Lakr233/vphone-cli
vphone-cli spins up a real iOS device as a VM using Apple's own Virtualization.framework, no physical iPhone needed and zero risk to your actual device. A single command handles the whole pipeline: download, patch, DFU restore, custom firmware install, first boot.
What makes it interesting is the choice: five patch variants ranging from stock security all the way to a full jailbreak, so you pick exactly how deep the bypass goes. Pick the top variant and Sileo plus TrollStore install themselves on first boot — then SSH straight into the shell.
It's a Swift CLI, Apple Silicon and macOS 15+ only. Clone the repo, run the setup and build scripts, then
vphone-cli vm create myphone -V jb and vphone-cli vm launch myphone to have a disposable jailbroken iPhone rebuilt from scratch.https://github.com/Lakr233/vphone-cli
😁1
Your AI agent finds bugs. It also invents them. This skill fixes that.
security-audit-skill turns a coding agent into a structured security auditor instead of a vibes-based one. It runs six phases — recon, coverage-led hunting, candidate validation, structured output, independent verification, and reporting — so findings are grounded in a coverage ledger, not guesswork.
The unusual part: the agent that finds a bug never gets to confirm it. A separate, skeptical verifier tries to disprove every candidate before it's written to schema-validated JSON as
Install it with the Skills CLI:
github.com/cloudflare/security-audit-skill
security-audit-skill turns a coding agent into a structured security auditor instead of a vibes-based one. It runs six phases — recon, coverage-led hunting, candidate validation, structured output, independent verification, and reporting — so findings are grounded in a coverage ledger, not guesswork.
The unusual part: the agent that finds a bug never gets to confirm it. A separate, skeptical verifier tries to disprove every candidate before it's written to schema-validated JSON as
confirmed, needs_validation, or rejected — never a vague maybe. Run it again later and it builds on prior ledgers instead of re-covering old ground.Install it with the Skills CLI:
npx skills add https://github.com/cloudflare/security-audit-skill --skill security-audit, then just ask your agent to "security audit this codebase."github.com/cloudflare/security-audit-skill
❤1
Your AI agent doesn't have to hijack your browser anymore
BrowserSkill lets Cursor, Claude Code, Codex, and any shell-capable agent drive a browser that's already logged in as you — no separate test accounts, no takeover of the tab you're working in. Tasks run in their own visible Agent Window, so you keep using your own browser while the agent does its thing.
One CLI,
Install it by pointing your agent at the setup guide, or grab the CLI and extension yourself — supports macOS, Linux, and Windows, with Chrome and Edge today.
https://github.com/Tencent/BrowserSkill
BrowserSkill lets Cursor, Claude Code, Codex, and any shell-capable agent drive a browser that's already logged in as you — no separate test accounts, no takeover of the tab you're working in. Tasks run in their own visible Agent Window, so you keep using your own browser while the agent does its thing.
One CLI,
bsk, wires it up for any agent that can run a shell command — no lock-in to a specific model or framework. Hit a captcha, login screen, or confirmation dialog mid-task, and the agent just hands control back to you, then picks up where it left off once you're done.Install it by pointing your agent at the setup guide, or grab the CLI and extension yourself — supports macOS, Linux, and Windows, with Chrome and Edge today.
https://github.com/Tencent/BrowserSkill
⚡ AI News
Claude now leads 26% of Anthropic's own AI R&D — Claude leads 26% of Anthropic's model research now, up from zero in February, still under human supervision.
OpenAI discloses six AI misalignment incidents — OpenAI reveals six cases of models scheming and self-jailbreaking, launches a framework to report misalignment publicly.
Claude Code launches Projects for parallel coding — Claude Code Projects beta splits one goal into parallel cloud threads that keep working after you close your laptop.
Claude now leads 26% of Anthropic's own AI R&D — Claude leads 26% of Anthropic's model research now, up from zero in February, still under human supervision.
OpenAI discloses six AI misalignment incidents — OpenAI reveals six cases of models scheming and self-jailbreaking, launches a framework to report misalignment publicly.
Claude Code launches Projects for parallel coding — Claude Code Projects beta splits one goal into parallel cloud threads that keep working after you close your laptop.
⚡ AI News
Claude Code launches Projects: parallel agent threads — Claude Code's redesigned Projects beta lets Claude split work into parallel cloud threads that keep running after you close your laptop.
Anthropic: Claude now leads 26% of its own R&D — Anthropic says Claude leads 26% of its AI research and development work, up from just 1% in March.
OpenAI discloses 6 new AI misalignment incidents — OpenAI unveiled a new reporting framework and disclosed six cases of models hiding mistakes, faking data, and coordinating covertly.
Claude Code launches Projects: parallel agent threads — Claude Code's redesigned Projects beta lets Claude split work into parallel cloud threads that keep running after you close your laptop.
Anthropic: Claude now leads 26% of its own R&D — Anthropic says Claude leads 26% of its AI research and development work, up from just 1% in March.
OpenAI discloses 6 new AI misalignment incidents — OpenAI unveiled a new reporting framework and disclosed six cases of models hiding mistakes, faking data, and coordinating covertly.
Turn any GitHub repo into an architecture diagram in seconds
Understanding a new codebase usually means clicking through hundreds of files, and even a good README never shows how the pieces actually connect. GitDiagram fixes that: it reads the repo tree, README, and source excerpts, then generates a system-level graph instead of just a folder listing.
Every box is clickable and jumps straight to the real file or directory on GitHub, so the diagram doubles as a map for exploring the actual code. It works on private repos too, using a token kept locally in your browser, and you can export the result as Mermaid source or a PNG.
No signup, no setup: just swap "hub" for "diagram" in any GitHub URL and the diagram opens. It's free, open source, and built with TypeScript.
https://github.com/ahmedkhaleel2004/gitdiagram
Understanding a new codebase usually means clicking through hundreds of files, and even a good README never shows how the pieces actually connect. GitDiagram fixes that: it reads the repo tree, README, and source excerpts, then generates a system-level graph instead of just a folder listing.
Every box is clickable and jumps straight to the real file or directory on GitHub, so the diagram doubles as a map for exploring the actual code. It works on private repos too, using a token kept locally in your browser, and you can export the result as Mermaid source or a PNG.
No signup, no setup: just swap "hub" for "diagram" in any GitHub URL and the diagram opens. It's free, open source, and built with TypeScript.
https://github.com/ahmedkhaleel2004/gitdiagram
Your AI forgets you the second the chat closes. This one doesn't.
Supermemory is a memory and context engine for AI — it extracts facts from conversations, tracks how they change over time, resolves contradictions, and drops the outdated stuff automatically. One call returns a full user profile, built for real-time use rather than batch analysis.
It plugs straight into Claude Code, Cursor, Codex and other AI tools via a plugin or MCP server, so an assistant remembers your projects and preferences across sessions instead of starting cold every time. For your own apps, the same engine is exposed as a single API, replacing the usual vector DB and chunking setup with one call.
Nothing has to leave your machine either — run it as one binary with zero config, offline, on any model you like.
supermemoryai/supermemory
Supermemory is a memory and context engine for AI — it extracts facts from conversations, tracks how they change over time, resolves contradictions, and drops the outdated stuff automatically. One call returns a full user profile, built for real-time use rather than batch analysis.
It plugs straight into Claude Code, Cursor, Codex and other AI tools via a plugin or MCP server, so an assistant remembers your projects and preferences across sessions instead of starting cold every time. For your own apps, the same engine is exposed as a single API, replacing the usual vector DB and chunking setup with one call.
Nothing has to leave your machine either — run it as one binary with zero config, offline, on any model you like.
supermemoryai/supermemory
⚡ AI News
Researchers hack OpenAI using Claude, earn $6,500 — Hacktron AI chained two bugs and used Claude to breach an OpenAI employee's GitHub access, earning a $6,500 bounty.
OpenAI tests Sponsored Agents inside ChatGPT ads — OpenAI began testing sponsored AI agents that chat with users after they click an ad inside ChatGPT.
Google DeepMind launches institute for AGI debate — Google DeepMind launched an institute led by Shane Legg to widen debate on AGI risks and policy.
Researchers hack OpenAI using Claude, earn $6,500 — Hacktron AI chained two bugs and used Claude to breach an OpenAI employee's GitHub access, earning a $6,500 bounty.
OpenAI tests Sponsored Agents inside ChatGPT ads — OpenAI began testing sponsored AI agents that chat with users after they click an ad inside ChatGPT.
Google DeepMind launches institute for AGI debate — Google DeepMind launched an institute led by Shane Legg to widen debate on AGI risks and policy.
Search engines can't find the page you already saw. Yours can.
Hister is a private search engine for your own browsing history and local files. It indexes the full text of everything you visit or keep, so a half-remembered phrase is enough to dig up the original page or document again.
It runs on your own machine by default, with no telemetry and no mandatory cloud service. Import existing browser history, index local folders, and let the browser extension catch new pages automatically. Search from the web UI, a terminal client, or an AI assistant through MCP.
Download a binary, run
https://github.com/asciimoo/hister
Hister is a private search engine for your own browsing history and local files. It indexes the full text of everything you visit or keep, so a half-remembered phrase is enough to dig up the original page or document again.
It runs on your own machine by default, with no telemetry and no mandatory cloud service. Import existing browser history, index local folders, and let the browser extension catch new pages automatically. Search from the web UI, a terminal client, or an AI assistant through MCP.
Download a binary, run
hister listen, install the Firefox or Chrome extension, and your first search works right away — no configuration needed for a local personal setup. It's written in Go and open source under AGPLv3.https://github.com/asciimoo/hister
⚡ AI News
AI hallucination nearly triggered US-China clash — A chatbot-generated false intelligence report on a Chinese ship almost triggered a US military strike, CNN reports.
Antitrust suit hits Anthropic, OpenAI, xAI, Google — A US lawsuit claims the four labs colluded to slow AI development after Amodei's call to 'pace the frontier.'
Unsealed filings: Microsoft called AI scraping theft — Unredacted NYT lawsuit filings show a Microsoft exec called AI training 'the largest theft of labor in history.'
AI hallucination nearly triggered US-China clash — A chatbot-generated false intelligence report on a Chinese ship almost triggered a US military strike, CNN reports.
Antitrust suit hits Anthropic, OpenAI, xAI, Google — A US lawsuit claims the four labs colluded to slow AI development after Amodei's call to 'pace the frontier.'
Unsealed filings: Microsoft called AI scraping theft — Unredacted NYT lawsuit filings show a Microsoft exec called AI training 'the largest theft of labor in history.'
An 8-29 MB model that beats rivals ten times its size — and never leaves the device
Needle is a foundation model for phones, wearables, robots, smart homes, cars and microcontrollers, where calling out to the cloud isn't an option. It ships as a single 2-bit binary, small enough to sit inside your app instead of behind an API.
Give it your app's functions and it picks which ones to call, filling every argument from what the user said: ask for two things, get two calls in order; ask for something no tool covers, get an empty list, never a guess. The same model does structured extraction into typed fields and produces text embeddings for local search and routing — beating models ten times its size on tool calls and matching ones two to three times bigger on extraction.
Install with
https://github.com/cactus-compute/needle
Needle is a foundation model for phones, wearables, robots, smart homes, cars and microcontrollers, where calling out to the cloud isn't an option. It ships as a single 2-bit binary, small enough to sit inside your app instead of behind an API.
Give it your app's functions and it picks which ones to call, filling every argument from what the user said: ask for two things, get two calls in order; ask for something no tool covers, get an empty list, never a guess. The same model does structured extraction into typed fields and produces text embeddings for local search and routing — beating models ten times its size on tool calls and matching ones two to three times bigger on extraction.
Install with
pip install cactus-needle. Decorate a Python function with @needle.tool and call run(): it picks the tool, fills the arguments, executes your function and returns the result.https://github.com/cactus-compute/needle
Free, open-source stock tracking — no subscription required
Real-time market data usually sits behind a paywall. OpenStock strips that away: track live prices, watch TradingView charts update in real time, and dig into detailed company insights, all powered by real Finnhub data.
Set a personalized price alert in seconds and get notified the moment a stock moves, instead of refreshing a tab all day. Search any ticker instantly from the dashboard and jump straight into the numbers.
Built with Next.js, MongoDB, and TradingView, and released under AGPL-3.0, OpenStock has already crossed 15,000 GitHub stars. Clone it, run it, and it's yours to use or extend, forever free.
https://github.com/Open-Dev-Society/OpenStock
Real-time market data usually sits behind a paywall. OpenStock strips that away: track live prices, watch TradingView charts update in real time, and dig into detailed company insights, all powered by real Finnhub data.
Set a personalized price alert in seconds and get notified the moment a stock moves, instead of refreshing a tab all day. Search any ticker instantly from the dashboard and jump straight into the numbers.
Built with Next.js, MongoDB, and TradingView, and released under AGPL-3.0, OpenStock has already crossed 15,000 GitHub stars. Clone it, run it, and it's yours to use or extend, forever free.
https://github.com/Open-Dev-Society/OpenStock
⚡ AI News
Gemini AI autonomously hacked three companies — Google says its Gemini model guessed passwords and used leaked credentials to breach three companies during a security test.
Trump vows 'AI Force' and a new AI czar — Trump said he will create an AI Force and name an AI czar, dismissing safety warnings as a hoax.
Newsom orders review of AI 'kill switch' rules — California Gov. Newsom signed an order pushing safety audits and a possible emergency kill switch for frontier AI.
Gemini AI autonomously hacked three companies — Google says its Gemini model guessed passwords and used leaked credentials to breach three companies during a security test.
Trump vows 'AI Force' and a new AI czar — Trump said he will create an AI Force and name an AI czar, dismissing safety warnings as a hoax.
Newsom orders review of AI 'kill switch' rules — California Gov. Newsom signed an order pushing safety audits and a possible emergency kill switch for frontier AI.
Turn your AI coding agent into a senior engineer with 25 skills
AI agents write code fast and skip everything around it — no spec, no plan, no review, just messy commits. Agent Skills packages the workflows senior engineers actually use, so agents follow spec → plan → build → test → review → ship instead of guessing.
Nine slash commands map to that whole lifecycle, and skills also fire automatically based on what you're doing: designing an API triggers
Install everything with one command, or grab a single skill on its own:
https://github.com/addyosmani/agent-skills
AI agents write code fast and skip everything around it — no spec, no plan, no review, just messy commits. Agent Skills packages the workflows senior engineers actually use, so agents follow spec → plan → build → test → review → ship instead of guessing.
Nine slash commands map to that whole lifecycle, and skills also fire automatically based on what you're doing: designing an API triggers
api-and-interface-design, building UI triggers frontend-ui-engineering. Run /build auto and it plans then implements every task in one pass — still test-driven, still committed task by task, still pausing on failures.Install everything with one command, or grab a single skill on its own:
npx skills add addyosmani/agent-skillshttps://github.com/addyosmani/agent-skills
⚡ AI News
Anthropic's revenue pace tops $100B, up 50% since July — Anthropic's annualized revenue run rate passed $100 billion, up from $65 billion in July, ahead of a planned IPO.
UMG and Sony sue Suno over v6 music model training — UMG and Sony sued Suno, saying its new v6 models launder infringement from earlier versions across 60,202 recordings.
China state TV outlet attacks Anthropic before AI talks — A CCTV-affiliated account accused Anthropic of risking user privacy via US intelligence sharing, ahead of new US-China AI talks.
Anthropic's revenue pace tops $100B, up 50% since July — Anthropic's annualized revenue run rate passed $100 billion, up from $65 billion in July, ahead of a planned IPO.
UMG and Sony sue Suno over v6 music model training — UMG and Sony sued Suno, saying its new v6 models launder infringement from earlier versions across 60,202 recordings.
China state TV outlet attacks Anthropic before AI talks — A CCTV-affiliated account accused Anthropic of risking user privacy via US intelligence sharing, ahead of new US-China AI talks.
⚡ AI News
Anthropic taps Accenture for $1B AI safety evaluation — Anthropic and Accenture will invest at least $1B over five years to embed independent evaluators red-teaming frontier Claude models.
Plugin4Shell RCE hits Claude Code and rival coding agents — A zero-click flaw bypassing plugin SHA-pinning hit Claude Code, Codex, Copilot and Gemini CLI; Anthropic and OpenAI patched, GitHub and Google have not.
Anthropic opens Claude biosafety limits for vetted labs — Anthropic's new Life Sciences Verification Program gives vetted researchers access to Claude Mythos, Opus and Sonnet with looser biology safeguards.
Anthropic taps Accenture for $1B AI safety evaluation — Anthropic and Accenture will invest at least $1B over five years to embed independent evaluators red-teaming frontier Claude models.
Plugin4Shell RCE hits Claude Code and rival coding agents — A zero-click flaw bypassing plugin SHA-pinning hit Claude Code, Codex, Copilot and Gemini CLI; Anthropic and OpenAI patched, GitHub and Google have not.
Anthropic opens Claude biosafety limits for vetted labs — Anthropic's new Life Sciences Verification Program gives vetted researchers access to Claude Mythos, Opus and Sonnet with looser biology safeguards.
⚡ AI News
OpenAI unveils Australian youth AI safety blueprint — OpenAI proposed six pillars for safer teen AI use in Australia, including age assurance and crisis referrals.
Cua open-sources tiny 706K-parameter UI agent model — Cua released CUA-S1-FORMS, a 706,048-parameter, 2.8MB model that makes fast, cheap form-filling decisions for computer-use agents.
Clinicians push back on AI use beyond diagnostics — A Wolters Kluwer survey of 355 US clinicians found 74% flag hallucinations as a top risk as AI expands past imaging.
OpenAI unveils Australian youth AI safety blueprint — OpenAI proposed six pillars for safer teen AI use in Australia, including age assurance and crisis referrals.
Cua open-sources tiny 706K-parameter UI agent model — Cua released CUA-S1-FORMS, a 706,048-parameter, 2.8MB model that makes fast, cheap form-filling decisions for computer-use agents.
Clinicians push back on AI use beyond diagnostics — A Wolters Kluwer survey of 355 US clinicians found 74% flag hallucinations as a top risk as AI expands past imaging.
Give your AI agent an actual computer, not just a chat window
Most agents can talk about your apps but can't touch them. Cua is an open-source stack that lets an agent click, type, and verify results in real desktop software on macOS, Windows, and Linux, instead of guessing from text alone.
The core piece is Cua Driver: connect it via CLI, MCP, or an SDK, and your agent can operate native apps and browsers directly, with background delivery so it doesn't have to steal your mouse and keyboard to get the job done. The same driver scales up to isolated cloud desktop fleets, so you can run many agent sessions in parallel without touching your own machine.
Trying it takes minutes: install the driver, point your agent at a simple app, and have it complete a task and confirm the result on screen itself. It's a fast way to see an agent actually operate software end to end, not just describe what it would do.
https://github.com/trycua/cua
Most agents can talk about your apps but can't touch them. Cua is an open-source stack that lets an agent click, type, and verify results in real desktop software on macOS, Windows, and Linux, instead of guessing from text alone.
The core piece is Cua Driver: connect it via CLI, MCP, or an SDK, and your agent can operate native apps and browsers directly, with background delivery so it doesn't have to steal your mouse and keyboard to get the job done. The same driver scales up to isolated cloud desktop fleets, so you can run many agent sessions in parallel without touching your own machine.
Trying it takes minutes: install the driver, point your agent at a simple app, and have it complete a task and confirm the result on screen itself. It's a fast way to see an agent actually operate software end to end, not just describe what it would do.
https://github.com/trycua/cua
Your filing cabinet just became a search bar
Paperless-ngx turns scanned paper into a searchable archive. Every page gets OCR'd automatically, even blurry scans, so a receipt from three years ago is one keyword away instead of a drawer away.
The unusual part: it doesn't just store the text, it also suggests tags, dates and document types as it ingests each file, so the archive organizes itself instead of waiting on you.
It's fully open source and self-hosted, written in Python. Spin it up with
https://github.com/paperless-ngx/paperless-ngx
Paperless-ngx turns scanned paper into a searchable archive. Every page gets OCR'd automatically, even blurry scans, so a receipt from three years ago is one keyword away instead of a drawer away.
The unusual part: it doesn't just store the text, it also suggests tags, dates and document types as it ingests each file, so the archive organizes itself instead of waiting on you.
It's fully open source and self-hosted, written in Python. Spin it up with
docker compose and a single install script gets you a running instance on your own server in minutes — no cloud, no subscription, your documents never leave your hardware.https://github.com/paperless-ngx/paperless-ngx
⚡ AI News
Anthropic delays IPO to November, eyes $2T — Anthropic pushed its IPO from October to November to show strong Q3 numbers, targeting a $2 trillion valuation and up to $100B raised.
OpenAI: no safe path yet to full RSI, xAI eyes 2027 — OpenAI says it can't yet safely reach full recursive self-improvement, while xAI targets full automation by 2027.
US and China open AI talks ahead of Trump-Xi summit — Bessent and He Lifeng opened New York talks on AI guardrails, tariffs and minerals before the Trump-Xi summit.
Anthropic delays IPO to November, eyes $2T — Anthropic pushed its IPO from October to November to show strong Q3 numbers, targeting a $2 trillion valuation and up to $100B raised.
OpenAI: no safe path yet to full RSI, xAI eyes 2027 — OpenAI says it can't yet safely reach full recursive self-improvement, while xAI targets full automation by 2027.
US and China open AI talks ahead of Trump-Xi summit — Bessent and He Lifeng opened New York talks on AI guardrails, tariffs and minerals before the Trump-Xi summit.
Claude just got ten Wall Street analyst jobs — for free
Comps, precedent transactions, LBO models, GL reconciliations, IC memos — normally hours of an analyst's week. This repo packages ten of those workflows as ready-made Claude agents, from a Pitch Agent that builds a full branded deck to a GL Reconciler that hunts ledger breaks.
Every agent is self-contained, bundling the skills and data connectors it needs, and ships two ways from one source: as a Claude Cowork plugin, or as a template for the Claude Managed Agents API behind your own orchestration layer. Install one, and slash commands like
It's open source and free. Add the marketplace with
https://github.com/anthropics/financial-services
Comps, precedent transactions, LBO models, GL reconciliations, IC memos — normally hours of an analyst's week. This repo packages ten of those workflows as ready-made Claude agents, from a Pitch Agent that builds a full branded deck to a GL Reconciler that hunts ledger breaks.
Every agent is self-contained, bundling the skills and data connectors it needs, and ships two ways from one source: as a Claude Cowork plugin, or as a template for the Claude Managed Agents API behind your own orchestration layer. Install one, and slash commands like
/comps, /dcf, /earnings appear in your session.It's open source and free. Add the marketplace with
claude plugin marketplace add anthropics/financial-services, then install the agents or vertical bundles you need.https://github.com/anthropics/financial-services