Every change you can see or use in every tool, newest first. Follow it in a feed reader with the Atom feed, or see what opens next in the release plan.
Sort by
Newest change
Name
Opening date
Release notes by tool
Every tool, its version.
13 of 18 tools are out and have release notes of their own; the rest list theirs from the day they open.
First release: build an AI character once, keep the same face everywhere
Character Prompt Builder writes image and video prompts for a character you describe step by step: identity, face, wardrobe, scene, camera and motion.
Seven kinds of character (a person, an animal, an anthro, an animated character, a robot, a creature or a fantasy race), each with its own questions.
99 ready-made characters in every kind and all 19 art styles: pick one and the builder opens with every answer filled in, ready to change, with its picture and name at the top of the live prompt for as long as the character itself stays the same. Every picture was generated with AI (Google Gemini) and says so with an AI mark.
Prompts come in three styles: plain sentences for Flux, Nano Banana, GPT Image, Midjourney v7, Firefly, Veo and Sora, Midjourney phrases with --ar and --v flags, or Stable Diffusion tags.
Every character comes with an identity block, the part you reuse word for word so the face stays the same, and a negative prompt tuned to its type.
Save characters to My characters, share one as a link or a character.json file, and load any of them again later.
It runs in your browser at /prompts/character-prompt-builder/app/, works offline once loaded, and needs no account or API key.
First release: find the right chart for your data by answering a few questions
Chart Compass picks a chart with you: paste your data, say what the chart should show, answer two or three questions, and see the charts that fit.
Paste a table from Excel or Google Sheets, open a CSV, or start from one of twelve samples; it finds the dates, categories and numbers by itself.
Eleven goals, from comparing and change over time to part of a whole and flow, with the ones your data can answer marked; all three steps fit on one screen. Paired questions never hold the same column: Actual and Target, and Across (X) and Up (Y), swap when you pick the other one's column.
The best fit and the good alternatives are drawn with your own numbers, each saying why it fits and which field goes where; on a phone they are drawn at the width they get, so their labels stay readable.
Any of the 50 charts can be the answer: it spots the data each one needs, such as two periods per item for a slope chart, daily dates for a calendar heatmap, start and end dates for a Gantt chart, or from and to columns for a Sankey or network.
Every pick names the chart in Power BI, Tableau and Excel, and the charts to skip for your data are listed apart, with why.
The chart map lays out 50 charts in eleven lanes, filtered by what you want to say, with a search that finds a chart by its name or its Power BI name; open one to see it drawn large, how to read it, a tip, the data it needs and similar charts.
Free, in your browser at /data/chart-compass/app/; your data never leaves the page. Your data, goal and answers are kept in this browser for your next visit, and Paste your own clears them.
First release: every Claude Code file, explained and written for you
Config Studio for Claude Code is an editor for the files Claude Code reads, from permissions and settings to CLAUDE.md, hooks, agents, skills, MCP servers and the status line: 20 surfaces in one list, checked against Claude Code 2.1.296 and its documentation, with all 388 environment variables, all 33 hook events and Claude Haiku 5.5 in the model lists and the cost estimate.
Each surface explains what it does and shows every file it can live in, the one that wins at the top, so you can see why a setting is ignored; a setting put in a file that does not read it is flagged, and every fact carries a badge saying where it came from.
Change an option and the exact file it would write updates beside it, in full and merged into what is already there, before anything touches your disk; Changes lists every key you set and the lines it writes.
Read in a setup opens the .claude folder you already have in the same editors and keeps what they cannot hold; credentials, caches, history and transcripts are never opened, and anything in a hook script worth a look is named with its line.
Start from one of 13 starter packs, check this machine so the plan warns about anything that will not run, and put rules, hooks, names and schedules to the test in the 21 engines under Try it.
Take the files, or one install script that merges them into your setup, keeps your old files in ~/.benibix-backups/ and writes an undo.sh that puts them back; or send the setup to somebody as a plugin, hooks included.
Save the plan to load it again later, or Start over to clear everything you set and everything this page kept in your browser.
It runs in your browser at /coding-agents/claude-code/config-studio/app/, with no account, and writes nothing to your machine until you run the script.
First release: every file Codex reads, explained and editable
Config Studio for Codex covers the 21 surfaces Codex reads, from config.toml, sandbox and approvals, permission profiles, rules, AGENTS.md and hooks to skills, MCP servers, custom agents, profiles, the status line and requirements.toml, checked against Codex 0.162.1 and its documentation.
Each surface shows every place it is read from, the one that wins at the top, and the exact file your choices produce; only what you set is written, and several surfaces that write config.toml come out as one valid file.
Read in a setup takes your .codex folder, a repository or any of their files and opens what is set there in the editors, one config.toml filling up to 12 surfaces; what no editor can hold is listed and kept, and your sign-in, sessions and history are never opened.
Pick and order the status line and terminal title items (31 footer items and 25 title items) and see the footer drawn the way Codex draws it.
Try it puts a command to the rules you type in: allowed, asked about or refused, with a chain of commands checked one by one.
Six starter packs: seatbelts, a read-only review profile, an unattended CI profile, a short list of allowed hosts, a useful footer and a team repository.
The install script merges into the config.toml you already have, keeping your trusted projects and every key you did not set, keeps your old files in ~/.benibix-backups/ with an undo.sh that puts them back, and prints the commands you built for later rather than starting Codex.
Save the plan to load it again later, or Start over to clear everything you set and everything this page kept in your browser.
First release: nearly 50,000 free icons, styled and exported from one place
Icon Studio searches eight of the best open-source icon sets together, Lucide, Tabler, Phosphor, Fluent, Heroicons, Bootstrap Icons, Iconoir and Material Symbols: nearly 50,000 icons in 23 styles, without company logos, whose owners do not allow recolouring them.
Search by name or by the words each set tags its icons with, or pick one of over 100 topics such as chart, money, user or calendar; a shared link, or the icon you had open last time, opens with that icon highlighted in the grid.
Style an icon with colour, size, stroke, rotation and flips, a solid or gradient background shape, a border, a shadow and CSS motion; Stroke reaches every outline icon, and a turned icon shrinks just enough that its corners stay in the frame.
A live preview, as big as it fits, shows the icon at 16 to 48 pixels too, so you know it still reads small; it stays in view with the Style and Export tabs while the controls below it scroll.
Take it away as SVG, PNG up to 4×, React, CSS, a data URI or a Power BI measure that shows it in a table.
Select many icons and download them in your style as SVGs or PNGs in one zip, or as one sprite; star the icons of a project and save a style as a preset. Reset brings back the default style, with Undo in case you change your mind.
Every set is MIT, ISC or Apache licensed and free for commercial work, with each licence shipped alongside (the Material Symbols one says the icons were changed, as Apache 2.0 asks); it all runs in your browser.
First release: 455 KPIs with their formulas, code and best charts
KPI Atlas holds 455 KPIs for 30 areas of a business, from sales, finance and HR to manufacturing, healthcare, banking, ESG and cybersecurity.
Every KPI has a plain definition, why it matters, its formula, its unit, which way is good, how often to track it, and the data you need.
Ready-to-copy DAX, SQL and Excel for every KPI; an on-time measure never counts an unfinished item, and a query that joins two tables adds each one up before it joins them.
The best charts for each KPI, ranked, each with the reason it fits and its Power BI visual, drawn with sample numbers shaped like the KPI; a waterfall ends on the KPI's example value, negative ones too.
A benchmark where a reliable one exists, the common mistakes, and related KPIs to read with it.
Every area opens with a starter dashboard: the KPIs most teams track, each in its best chart.
Star KPIs into your set and take them all at once: every DAX measure, the SQL, a spreadsheet, a spec document or a link to share. Clear offers Undo.
Each area and each KPI has its own page and address, at /data/kpi-atlas/app/, and on a phone the area you are in stays in view and a KPI opens over the whole screen.
First release: 150 motion graphics, shaped live and taken away as code or a file
Motion Studio plays 150 ready-made motions at once, in ten kinds: loaders, text, backgrounds, buttons and hover, attention, entrances, UI feedback, shapes and SVG, charts and numbers, and pointer.
Open one to change its colours (with palettes that go together), size, speed, text and easing, and see the change as you drag.
Star the motions you like and find them under Favourites on the app bar, or copy a link that opens a motion with your settings; the motion travels in the part after #, which browsers do not send to the server.
Pause, drag to any moment, slow it down, and try it on a dark, light, see-through or any coloured stage. Hover and click motions are pressed for you, or you can try them yourself.
Pointer motions, such as a 3D tilt card, a magnetic button and watching eyes, follow a drawn path or your own pointer, and come with their few lines of script.
Shape the easing curve by dragging its handles, or pick one of 14, springs and bounces included.
Take the code as HTML and CSS, React, Vue or Svelte, or a page that plays it. Every motion stops by itself for people who ask their system for less motion.
Save it as a GIF, an MP4 or WebM video, an SVG that animates by itself, or the frame on the stage as a PNG, all made in your browser; long loops are recorded for one whole cycle.
First release: build a GitHub profile README with a preview that matches GitHub
Profile README Studio builds the README at the top of your GitHub profile from 45 block types and 31 layouts (2 for organizations), with a preview that follows github.com's own styles in light and dark, at desktop or phone width.
Click anything on the page to edit it, type straight into headings, add a block in any of its ready-made styles, and reorder blocks by drag or keyboard in the Outline.
The studio draws every badge, logo tile, banner, divider, callout title, typing line (with its font inside) and tech stack icon itself. They become files in a profile-assets folder of your profile repository: Export downloads them all as a .zip, Publish for me commits them with a token you create.
Stats, top languages, streak, trophies, the activity graph, featured repos, gists, the contributors wall, the contribution snake, the 3D contribution graph, a Year overview and WakaTime are drawn by one workflow, profile.yml, in your own repository, with only GitHub's own actions. It also writes your recent activity and latest blog posts into the README, and an optional PROFILE_TOKEN secret adds your private commits, pull requests and issues as numbers.
The preview loads nothing from other sites: GitHub's own pictures appear once you enter a username, and Check pictures confirms that every picture is in your profile repository and says what to do about a missing one.
Fill blocks from your public GitHub profile or from a resume PDF read in your browser, import the README you already have (known widgets from public servers become the studio's own blocks), and share your layout as a link.
The Profile Doctor scores any profile README out of 100 on 9 rules and opens a fixed version next to the original. Write with AI builds a prompt for any assistant, or writes the text with your own Anthropic key, kept for the tab only.
57 widget themes, with an Auto theme that gives light and dark visitors the right card, and a free course of 12 lessons, from your first README to an organization profile.
First release: tested image prompts, each next to the image it made
Prompt Gallery is a library of 367 image prompts for Midjourney, Flux, Nano Banana and more, in 8 categories and 54 subcategories, among them 13 character sheets.
Every prompt sits next to the real image it produced, so you see the result before you copy anything. Every example image was generated with AI (Google Gemini) and says so with an AI mark, and a downloaded image's name ends in -ai-generated.
Search titles, prompts, tags and model names, filter by subcategory, and sort by featured, newest, name or subcategory. An open prompt lists the models it was written for; click one to see all its prompts.
Customize a prompt before you copy it: fill in its placeholders, add your own details, pick an aspect ratio and apply a saved brand kit.
7 kits group the prompts for one job, such as a pitch deck, a product launch or an annual report.
Favourites stay in your browser and move between devices as a file, every view has a link to share, and the whole library is open as JSON at api/prompts.json.
It runs in your browser at /prompts/prompt-gallery/app/, with no account.
First release: design a Power BI theme and a report page, live
Report Design Studio builds a Power BI theme on a live report preview that changes as you edit, and designs the page around it. Undo and Redo step through your changes (Ctrl+Z, Ctrl+Shift+Z), and Reset goes back to the report you first saw.
Start from 65 curated palettes, colour-blind-safe sets such as Okabe–Ito and Paul Tol among them, generate one with harmony rules, extract one from an image, or build your own, with sequential and diverging ramps and scientific colormaps.
Simple mode styles every visual at once; Advanced mode styles one of 27 visuals at a time, buttons and shapes also in their hover, selected and pressed states, plus the report wallpaper and the filter pane.
Layout designs the whole page: navigation (an icon rail, a sidebar, a floating sidebar or a top bar), a title bar, a slicer panel, sections of tiles and a footer, from one of eight starters, at Power BI, Tableau, Looker Studio or Full HD sizes.
One Accessibility panel checks every colour against its card and the 3:1 floor for chart marks, shows the report with colour blindness, and fixes contrast in one click while keeping the colours apart.
Load an existing theme JSON to edit it: it is validated on the way in, its colours stay exactly as they are and in their order, and settings the studio does not edit are kept in the export.
Export the theme JSON, a matched dark and light pair, a swatch sheet, design tokens for CSS, SCSS or Figma, and the page as a PNG, a layered SVG or a PowerPoint slide of editable shapes.
It runs in your browser with no account. A share link holds the whole design in the part of the address a browser never sends to a server, and presets and layouts stay in your browser; every panel works from the keyboard.
First release: build a Claude Code status line without writing a script
Statusline Studio builds the status line under Claude Code's prompt: pick segments, style them, and watch the line render live before you install it.
75 segments for the status line and 15 for subagent rows, from model and cost to git branch, context use, plan limits, Spend in dollars and prompt cache misses, and 25 templates to start from. Subagent rows show each subagent's type, such as Explore or your own code-reviewer.
Every segment comes as text, an emoji or a Nerd Font icon, and one switch changes all of them at once; the preview draws the icons even without a Nerd Font installed, and the icon picker takes your own.
Edit right in the preview: click a segment to change it and drag it to move it. Undo and Redo in the bar (Ctrl+Z, Ctrl+Shift+Z) take back or bring back any change, a Reset too. In the layout, the arrow keys move a segment along its line or to another line.
Preview data shows your line in a fresh, working, heavy or near-the-limit session, and segments with nothing to show, such as the limits for an API key, hide instead of printing dashes.
Export the script in Bash, Python or Node.js, or copy one install command that saves it and turns it on in settings.json. It backs up your old settings first, and leaves a settings.json with a typo as it was, printing the lines to add by hand.
Share a design as a link; a shared design that runs shell commands shows you each one and asks before keeping it, and every option is checked before it reaches the script.
A Not working? guide with a one-paste self-check covers the common problems, from a missing jq to a folder you have not trusted yet, and it is free in your browser at /coding-agents/claude-code/statusline-studio/app/, with no account.
First release: see what you did with Claude Code, from your own history
Usage Insights for Claude Code turns the history Claude Code keeps in ~/.claude/projects into a dashboard, read in your browser, with nothing uploaded.
Me shows your rhythm: active days, streaks, a weekday by hour heatmap, projects, languages, lines changed, commits and how many of Claude's edits you kept, plus cost per day and by model (each model once, its 1M-context variant counted with it), session length and your habits, such as tool calls per prompt and prompts outside working hours.
Capabilities lists every skill with its plugin, every MCP server down to each of its tools, every built-in tool, subagent and slash command, with calls, success, sessions and last use, and marks the ones used for the first time.
Team adds everyone's exported events for active people against seats, retention and adoption by team, with teams under five people merged or hidden.
Telemetry answers whether it is working: turn duration p50 and p95, where the time goes between the model, tools and retries, API errors, permission decisions, connector health, compactions and agent runs, with the average time and tools of subagents that run in the background.
The Catalog lists all 73 metrics with their definition, source and chart, and a sample team lets you try everything before loading your own data.
Export the whole dashboard as one HTML file that opens offline or as a PDF on A4 landscape pages, your events in the unified schema with a pseudonymous id, or a daily CSV for a spreadsheet or BI tool.
A free course of seven lessons, from opening your history to a team dashboard and the metrics catalog.
First release: see what you did with Codex, from your own session files
Usage Insights for Codex turns the session files Codex keeps in ~/.codex/sessions into a dashboard, read in your browser, with nothing uploaded; compressed sessions (.jsonl.zst) are read too, and auth.json and config files are skipped even inside a dropped folder.
Me shows your rhythm: active days, streaks, a weekday by hour heatmap, projects and branches, languages patched, lines changed, commits and how many of Codex's patches you kept.
Usage limits shows the highest share of your 5-hour and weekly windows used each day, the peaks and how often a limit was reached, from the limits Codex records with every token count, beside your tokens, reasoning and models.
Capabilities lists every skill and who started it, every MCP server down to each tool, the built-in tools and each subagent with its time and tool calls.
Telemetry answers whether it is working: turn duration and time to first token at p50 and p95, turn errors by class, command and patch success, sandbox and approval modes, connector health and compactions.
Team adds everyone's exported events for active people against seats, retention and adoption by team, with teams under five people merged or hidden.
The Catalog lists all 70 metrics with their definition, where Codex records them and their chart, and a sample team lets you try everything before loading your own data.
Export the whole dashboard as an HTML file that opens offline or as a PDF, your events in the unified schema with a pseudonymous id and no thread names, a summary of every number, or a daily CSV.