Obsidian is a free notes app that saves every note as a plain Markdown file in a folder on your own machine, then lets you link those notes together. 👉 https://obsidian.md How it works: 1. Pick a folder. Obsidian calls it a vault. It can be a new empty folder or one you already have. 2. Write notes. Each one is saved as a .md text file in that folder, so other text editors can open it too. 3. Link them. Type [[ and pick a note to link to it. Rename a note later and Obsidian can update the links to it. 4. Follow the connections. The Backlinks panel lists every note that links to the one you're on. Graph view draws the whole vault as circles joined by lines. The more notes point at one, the bigger its circle gets. That plain folder is where Claude Code fits. Open Claude Code inside the vault and it reads and writes your notes like any other files. Obsidian refreshes by itself when something outside the app changes them. Obsidian also has a command line tool. Switch on Command line interface in Settings under General, then follow the prompt to register it. Run obsidian search to find text across the vault, or obsidian backlinks to list what links to a note. Claude Code can run both like any other terminal command. It needs installer 1.12.7 or newer, with the app running. On cost: the app is free with no sign-up, even for work (an optional commercial licence is $50 per user a year). Sync is $4 per user a month and Publish is $8 a month per site, both billed yearly. Could Claude Code read your notes right now, or are they locked inside an app?

REA (Reverse Engineer Anything) has been all over GitHub and X. In this video I show what it actually does, then use it with Claude Code to rebuild the one Wispr Flow feature I use every day: hold a key, talk, let go and the text lands wherever your cursor is. Step 1, What REA is: REA is an MCP server. Claude Code asks it to look at an app, and it runs tools like Ghidra, a JavaScript analyser, Chrome and JADX to take the app apart. Claude gets the evidence back and either explains how the app works or builds its own version. Step 2, Setup: One command in your terminal, npx rea-agents setup. Pick Claude Code, then restart it. Step 3, The prompt: One prompt, dictation only. REA reads Wispr Flow's files without ever launching the app and labels every finding observed, inferred or unknown. Step 4, The test: About 15 minutes later Claude came back with Hold to Type. Hold Right Option, speak and release. The text pastes wherever your cursor is. It runs on a Whisper model on my Mac, so nothing goes to the cloud. Step 5, Is this new? Not really. The NSA made Ghidra free back in 2019 and these tools have been around for years. What REA adds is all of them behind one connection your agent can use, so Claude works from evidence instead of guessing. The prompt I used: Use REA to work out how Wispr Flow's dictation works: the hotkey you hold, the recording, the speech-to-text, and how the text gets typed wherever the cursor is. It's installed on this Mac. Read its files only, don't launch it. Label every finding observed, inferred or unknown. Then build my own version of just that: hold a key, speak, let go, and the text is typed wherever my cursor is. Use the local Whisper model on this Mac, nothing in the cloud. Resources: REA on GitHub → https://github.com/morluto/rea REA website → https://rea.tools/

Google's new top model went on general release on 13 October. If you build on Claude, the price is the part worth knowing. https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-argon/ Until 30 November it costs $2 per million input tokens and $10 per million output. That is exactly what Claude Sonnet 5.5 costs. On 1 December it doubles to $4 and $20, which is exactly Claude Opus 5.5. Cached input matches too: $0.10 now, $0.20 from December. Where you can use it: 1. Google AI Studio, on the paid Gemini API tier or a Google AI Ultra plan. There is no free tier for Argon. 2. The Gemini app, on Google AI Pro or Ultra. 3. Antigravity, Google's coding tool, on Ultra. How it stacks up against Claude, from VentureBeat's launch write-up: - Vals AI's index of finance, legal, coding and tax work: Argon 68.90%, Sonnet 5.5 67.04%, Opus 5.5 66.97%. - Terminal-Bench, the test closest to Claude Code style work: Opus 5.5 66.4%, Argon 57.4%. Two things to know before you test it. Thinking runs at low, medium or high. And the 1 million token output in the headlines only applies with Google's long decode mode, so normal replies are far shorter. The quickest way to decide: take one task you already run on Claude, give Argon the same prompt in AI Studio, then compare the output and the token count. Until 30 November it costs the same as Sonnet, so the test is cheap. Sources: https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-argon/ https://ai.google.dev/gemini-api/docs/pricing https://claude.com/pricing https://venturebeat.com/technology/google-makes-gemini-4-argon-broadly-available-with-limited-time-low-api-pricing-of-2-10-per-1m-token

Archify turns a sentence or a whole codebase into an interactive diagram you can click through. It has 83,740 stars on GitHub. https://github.com/tt-a1i/archify It works in Claude Code, Codex, Cursor and OpenCode. Here's what happens when you ask it for a diagram: 1. You describe the thing. One line is enough: Browser, API, Redis cache, PostgreSQL fallback. Or open a repo and ask it to map the runtime architecture. 2. Claude writes the diagram as a typed JSON source file first. 3. Bundled validators check the layout, routes and labels. If a check fails, it gets back the exact rule that broke and the fixes allowed. 4. It renders the HTML and checks it again. Only a version that passes replaces the file you see. 5. You refine it in chat. Add Redis, move auth to the left, highlight the rollback path. The rest of the diagram stays where it was. There are ten diagram types: architecture, workflow, sequence, data flow, lifecycle, ERD, class, tree, timeline and waterfall. The last five landed in version 3.1.0 yesterday. The output is one HTML file. Open it in any browser, press / to find a node, R to trace a path, F for presentation mode. Send it to anyone and the clicks still work. It also exports PNGs and share cards. Install it with: npx skills add tt-a1i/archify -g See the live demos first: https://archify.si/gallery What would you map first, your own stack or a client's workflow? Drop the diagram in the comments.

Anthropic just let us mod Claude Code. A mod changes what you see and how Claude Code behaves: panes, buttons and live info right above the prompt. You don't write the code yourself. You describe what you want and Claude builds it. In this video I show five I built. All five are below for you to install. The five: Chomp: A maze game opens beside the chat once Claude has been working for 20 seconds. A line above the maze shows what Claude is doing. The game pauses when it's done. Type /chomp to open it any time. Cache Clock: One line above the prompt: minutes left before your cache goes cold, what your next message would cost if it does, your context, your five-hour and weekly limits and the session cost. Handoff makes Claude write a note so a fresh chat picks up where you left off. Command Centre: Type /centre. Press Find my prompts and it reads your recent sessions, finds the prompts you repeat most and turns them into one-press buttons. The Skills tab puts every skill and command one press away. Prompt Coach: Type /coach and your prompt. It scores it out of 10 on goal, context and done, tells you what's missing and gives you a better version to send or edit. Model Showdown: Type /versus and a question. It goes to Haiku, Sonnet and Opus at once and shows the answers side by side with time, tokens and cost. Use this one sends the answer you pick into your chat. Before you install: You need Claude Code v2.1.287 or later in the terminal, or the Code tab of the Claude Desktop app on v2.1.286 or later. Type /status to check. Mods run in VS Code, but nothing shows in its chat panel. Install in your terminal, two lines per mod: claude plugin marketplace add automatorsplus/chomp-mod claude plugin install chomp@chomp-mod claude plugin marketplace add automatorsplus/cache-clock-mod claude plugin install cache-clock@cache-clock-mod claude plugin marketplace add automatorsplus/command-centre-mod claude plugin install command-centre@command-centre-mod claude plugin marketplace add automatorsplus/prompt-coach-mod