dgrabovets-plugins
Ponytail mode and skills plus one band above the Claude Code prompt.
Safety scan
A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.
Check it deeper on your machine
It has changed since its scan on 20 Nov 2026.
As a Claude Code plugin, install it and then ask Claude Code:
/plugin marketplace add keithmackay/mackayi/plugin install scanskillsafety@mackayiDo a deep scan of https://github.com/Dgrabovets/UnifiedBand with scanskillsafety
Or from a terminal, clone it and run it:
git clone https://github.com/keithmackay/scanskillsafety-skill ~/.claude/skills/scanskillsafetynode ~/.claude/skills/scanskillsafety/cli.cjs --deep 'https://github.com/Dgrabovets/UnifiedBand'
It checks the exact commit: every file, its packages against OSV (api.osv.dev), how it compares with this page's scan (findsafeskills.com receives the repository's address), and, if you agree, Claude reading every file with your own Claude subscription. It never runs the code.
Badge for your README
Markdown
[](/listings/dgrabovets--unifiedband)HTML
<a href="/listings/dgrabovets--unifiedband"><img src="/api/listings/dgrabovets--unifiedband/badge.svg" alt="findsafeskills safety rating" /></a>Install
Install instructions from the README
As written by the repo's author. findsafeskills has not verified these steps.
/plugin marketplace add Dgrabovets/UnifiedBand /plugin install unified-band@dgrabovets-plugins
Built and tested on Claude Code 2.1.302. The band above the prompt needs a build with mod support; without it only the Ponytail hooks and skills load.
What was scanned
The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.
.claude-plugin/marketplace.json · Manifest · 16 lines
- {
- "$schema": "https://anthropic.com/claude-code/marketplace.schema.json",
- "name": "dgrabovets-plugins",
- "description": "Ponytail mode and skills plus one band above the Claude Code prompt.",
- "owner": {
- "name": "dgrabovets"
- },
- "plugins": [
- {
- "name": "unified-band",
- "description": "Ponytail mode, skills and task tracking in one mod, with a single band above the prompt: controls, session info, cost, tests, limits, model router and project completion.",
- "source": "./",
- "category": "productivity"
- }
- ]
- }
README.md · README · 86 lines
- # Unified Band
- [](https://github.com/Dgrabovets/UnifiedBand/actions/workflows/ci.yml) [](LICENSE)
- Ponytail (mode, skills, task tracking and board) merged with one band above the Claude Code prompt, in a single mod.
- ## What it shows
- - Level buttons: Off, Lite, Full, Ultra (the current level is highlighted).
- - Action buttons: Review, Audit, Board, Compact.
- - Two info lines: folder, branch, session age, prompt count, cost; last turn duration, model, cache hit rate, idle time, compactions.
- - Cards that stretch to fill each row: Branch, Cost, Tokens, Agent Time, Reuse, Callers, Tests, Context, 5-Hour Limit, Weekly Limit, Project Completion.
- - Toasts when a usage limit crosses 80% and 95%.
- Project Completion is the share of checked boxes in a Markdown checklist (`- [x]` over all `- [ ]` and `- [x]` lines). The file is `PLAN.md` at the repo root; change it with the `progressFile` option. The card is hidden when the file is missing.
- ## Requirements
- - Claude Code with mod support (hooks modules).
- - Node.js on PATH (the mode hooks are Node scripts).
- - Disable the original Ponytail plugin: this mod contains its code, and both together double every hook.
- ## Credits
- The mode, skills, hooks, task tracking and board come from [Ponytail](https://github.com/DietrichGebert/PonyTail) by DietrichGebert (MIT, see `LICENSE-PONYTAIL`). Changes: the state key and skill namespace are `unified-band`, the band is redrawn, and the mode tracker also answers `/unified-band:ponytail`.
- ## Board
- The Board button opens a local page (no server, no network) built from three sources: the task ledger, the event journal and the model router's log. The ledger is one CSV per project and session (`<project>-<session>.csv`, and `<project>-import.csv` from `/ponytail band import`); the Board merges them.
- - **Share card** is a poster: the session count as the big number, a line with projects, cost and agent time, four figures (active days, streak, reuse, plan pace) and two bar panels (cost and tokens per day of the chosen period) along the bottom. It is shown whole, scaled to fit.
- - **Overview**: four cards with sparklines (Sessions, Agent time, Cost, Tokens), a stacked chart by project, the Activity honeycomb, period 7d, 30d, 90d, All, and the Share and Export buttons: Share holds Share card and Copy summary, Export opens a dialog: pick sections and options on the left, a format (PDF, Markdown, Raw data) on the right, then Download.
- - **Projects** (on the Overview page, above Activity) is one table that is also the chart's legend: a swatch shows or hides a project in the chart, filters pick a category, Large projects open their own page. With a project chosen it shrinks to one line. Every project worked on, in four categories (Large, Medium, Small, Chat) scored from turns, cost, sessions, days of work and whether it has a `PLAN.md`. Click a Large project for its own page with all the sections below filtered to it.
- - **Context and limits**: context fill curves, fill rate, life before compaction, 5-hour window peaks, burn rate, weekly forecast, blocks.
- - **Reuse**, **Model Router** (matrix from your model to the chosen one, savings estimate, last decisions), **Plans** (burn-up, velocity, projects without a plan), **Review and Audit**, **Extras** (cache, tests, cost by model, working hours, expensive sessions, push backs, subagents).
- - **Motion**, always on: pages fade in, cards rise in turn, numbers roll like a departures board, lines and charts draw in, the main chart morphs between metrics, the honeycomb pops in as a wave and the hottest cells glow, bars and columns grow, a limit close to its end breathes a red ring, a streak of a week or more throws confetti once, the share card assembles itself, cards glow under the pointer. Only transform and opacity move (plus the clip of a drawing chart and the shape of the morphing one). The system "reduce motion" setting turns all of it off.
- - **Project switcher** in the header on every page: all projects, or one of the large ones.
- - **Raw data**: one JSON with the ledger, the journal, the router log, the config and every section's numbers, with a schema, for sending to someone who will analyse it. It follows the sections ticked in the Export dialog (the ledger, the journal and the router log are included only when a ticked section uses them). Session titles only if you tick "Include session titles"; "Hide project names" masks the names.
- - **Pages**: the sections are tabs, one page at a time. Ad hoc chats (the app's `spare-*` and `scratch-*` folders, `sandbox`, a drive root) are one group, "Chats", counted in sessions and never opened per chat.
- - **Export**: every section has Export (PDF, Markdown, CSV). PDF opens the browser's print dialog: choose "Save as PDF". Export all takes any sections, with an option to hide project names. Copy summary puts a short Markdown report on the clipboard.
- The event journal lives in `~/.config/ponytail/band/events-YYYY-MM-DD-<session>.jsonl` (one file per session and day): one line per turn, prompt, compaction, `/model` command, subagent turn and review or audit run. It never stores prompt text, only counts and flags. History starts the day the journal is first written.
- The dashboard is called Unified Band, with a hexagon-and-bars mark of its own. Set `"title"` in `board.json` to call it something else; the name shows in the sidebar, the tab, the share card and the exports.
- ### board.json
- Optional, next to the journal. Rename a project, force its category, merge two folders into one project (`aliases`) or add folders to the Chats group (`chats`):
- ```json
- {
- "projects": { "SipAndTell": { "label": "Sip & Tell", "category": "Large" } },
- "aliases": { "unified-band": "UnifiedBand" },
- "chats": ["sandbox2"]
- }
- ```
- ### Checking the page
- `node scripts/smoke-board.mjs out.html` builds the page from your real ledger plus a synthetic journal and runs its script against a stand-in DOM. Needs Node 22.15 or later.
- ## Install
- ```
- /plugin marketplace add Dgrabovets/UnifiedBand
- /plugin install unified-band@dgrabovets-plugins
- ```
- Built and tested on Claude Code 2.1.302. The band above the prompt needs a build with mod support; without it only the Ponytail hooks and skills load.
- ## Known limits
- - The panel covers Ponytail's own row only after it has drawn once. If you see two panels after a reload, the cover hook did not run on your build.
- - Buttons are text only: the `Button` element has no icon prop.
- - The Board button builds and opens `~/.config/ponytail/band/board.html` from what has been recorded so far; before the first finished turn it shows an empty page.
- - Type-checking with `tsc` was not run.
- ## Test
- CI runs the same checks on every pull request and every push to `master` (`.github/workflows/ci.yml`): validate, the tests, and the Board smoke test.
- ```
- claude plugin validate .
- claude plugin test .
- ```
scripts/router-corpus.mjs · Bundled script · 156 lines
- // Prompts for the router differential test: [name, prompt].
- // Every prompt runs under the default parent (no transcript, so Opus) and under a Sonnet session;
- // `extra` lists more cases for a prompt: { parent, plan, ctx, transcript, config }.
- // After a change here run: node scripts/router-differential.mjs <original plugin folder>
- const long = 'Please walk through the billing module and explain how invoices are generated, how taxes are applied per region, how refunds are reconciled with the ledger, how the retry policy for failed payments works, which background jobs touch the invoice table, and where the currency conversion rates are cached and refreshed so that I can write accurate documentation for the new support team'
- const prompts = [
- // short lookups
- ['lookup-what-time', 'what time is it?'],
- ['lookup-list-prs', 'list open PRs'],
- ['lookup-show-log', 'show me the git log'],
- ['lookup-where-config', 'where is the config loaded?'],
- ['lookup-find-refs', 'find references to FooBar'],
- ['lookup-status', 'status?'],
- ['lookup-name-with-arg', 'name the three largest files in src'],
- ['lookup-name-bare', 'name'],
- ['lookup-long-line', 'show me ' + 'every single file '.repeat(10) + 'in the repository'],
- // complex verbs
- ['complex-short', 'refactor the rate limiter'],
- ['complex-migration', 'please migrate the user table to Postgres'],
- ['complex-code', 'refactor this function:\n```python\ndef foo(x):\n if x: return 1\n return 0\n```\nto remove the branch'],
- ['complex-breadth', 'redesign the whole architecture'],
- ['complex-breadth-word', 'port the codebase to ESM'],
- ['complex-many-words', 'implement a retry wrapper around the HTTP client so that every outbound request backs off exponentially and respects the Retry-After header'],
- ['complex-multiline', 'rewrite the loader\nkeep the API\nadd tests\nupdate docs'],
- ['complex-optimize-uk', 'optimise the query planner'],
- ['complex-extract', 'extract the parser into its own package'],
- // deep verbs
- ['deep-plan', 'plan the rollout of the new auth flow'],
- ['deep-design-paths', 'design a new auth flow that supports SSO, magic links, and passkeys across api/auth.py, ui/login.tsx, and tests/test_auth.py, include the migration story and a rollback plan for the sessions table and the cookie format we ship today'],
- ['deep-audit', 'audit the dependencies'],
- ['deep-investigate-q', 'investigate why the nightly job is slow?'],
- ['deep-root-cause', 'find the root-cause of the flaky login test in ./tests/login.spec.ts and ../ci/run.sh'],
- ['deep-profile-uppercase', 'PROFILE the importer'],
- ['deep-word-inside', 'the planner is broken'],
- ['deep-hyphen', 'do a root cause analysis'],
- // standard
- ['standard-fix', 'fix the typo in the README title'],
- ['standard-five-words', 'update the changelog entry today'],
- ['standard-four-words', 'fix the login bug'],
- ['standard-long', long],
- ['standard-too-long', long + ' ' + long],
- ['standard-two-lines', 'fix the off-by-one in the paginator\nit shows page 0 instead of 1'],
- ['standard-six-lines', 'a\nb\nc\nd\ne\nf is the failing letter in this alphabet test'],
- ['standard-code-block', 'why does this fail?\n```js\nconst a = 1\n```'],
- ['standard-test-verify', 'please verify the build works after the change to the config loader'],
- ['standard-question-short', 'is this thread safe?'],
- ['standard-tab-separated', 'fix\tthe\tflaky\ttest\tplease'],
- ['standard-nbsp', 'fix the flaky test please'],
- ['standard-crlf', 'fix the bug\r\nin the parser\r\nplease today'],
- // fan-out
- ['fanout-numbered', 'do these:\n1. rename the config key\n2. update the docs\n3. bump the version'],
- ['fanout-numbered-paren', '1) add a flag\n2) wire it up'],
- ['fanout-bullets', 'tasks:\n- add a header\n- fix the footer\n- trim the sidebar'],
- ['fanout-two-bullets', 'tasks:\n- add a header\n- fix the footer'],
- ['fanout-for-each', 'update the license header for each module in the repo'],
- ['fanout-for-all-of', 'run the linter for all of the packages'],
- ['fanout-verbs', 'add a test and fix the bug and update the docs'],
- ['fanout-verbs-semicolon', 'add a test; fix the bug; update the docs'],
- ['fanout-two-verbs', 'add a test and fix the bug'],
- ['fanout-fullwidth-digits', '1. rename it\n2. move it'],
- ['fanout-bullet-char', '• add a header\n• fix the footer\n• trim the sidebar'],
- ['fanout-plus-refactor', 'refactor the api:\n1. split the router\n2. move the handlers'],
- // session flow, overrides, opt-outs
- ['followup-yes', 'yes'],
- ['followup-ok-go', 'ok go ahead with that'],
- ['followup-continue', 'continue'],
- ['followup-also', 'also fix the tests'],
- ['followup-too-long', 'yes please do that and then also the other thing we talked about earlier today'],
- ['followup-override', 'yes #model=haiku'],
- ['override-haiku', 'refactor the whole parser #model=haiku'],
- ['override-sonnet', 'design the cache #model=sonnet'],
- ['override-opus', 'do the thing #model=opus'],
- ['override-fable', 'write the migration #model=fable'],
- ['override-upper', 'do the thing #MODEL=OPUS'],
- ['override-suffix', 'do the thing #model=opusx'],
- ['noshift', 'refactor the parser #noshift'],
- ['noroute', 'refactor the parser #noroute'],
- ['harness-task', '<task-notification>done</task-notification> refactor the parser'],
- ['harness-reminder', '<system-reminder> <scheduled-task name="x"/> plan the work'],
- ['harness-system', '[SYSTEM NOTIFICATION] build finished, now refactor the parser'],
- ['ultracode', 'run ultracode on the repo'],
- ['ultracode-inside-word', 'my ultracodes are fine'],
- ['goal-prefix', '/goal ship the release'],
- ['goal-driven', 'please run /goal-driven for this'],
- ['too-short-two', 'fix it'],
- ['too-short-question', 'why?'],
- ['too-short-one-word', 'hello'],
- ['empty', ''],
- ['whitespace', ' \n '],
- // other languages and odd characters
- ['ru-refactor', 'Перепиши модуль авторизации и добавь тесты для нового кода в auth.py'],
- ['ru-glued-verb', 'исправьplan баг в парсере сейчас'],
- ['ru-question', 'где лежит конфиг?'],
- ['ru-path-glued', 'смотрифайл src/main.ts и исправь ошибку в нем'],
- ['zh-standard', '请帮我修改这个函数的逻辑 并且 添加 注释 说明 这个 函数 的 用途'],
- ['zh-no-spaces', '请帮我重构这个模块并添加测试'],
- ['ja-standard', 'この関数を refactor してテストを 追加して ください'],
- ['de-standard', 'Bitte überprüfe den Code und behebe den Fehler in der Anmeldung heute noch'],
- ['de-deep', 'Wir müssen das Login-System designändern und dabei die Sicherheit prüfen'],
- ['ar-standard', 'مرحبا كيف يمكنني إصلاح هذا الخطأ في الكود الخاص بي اليوم'],
- ['mixed-translit', 'refactor-ть модуль пожалуйста сейчас'],
- ['emoji', 'please fix the build 🚀🚀🚀 it fails on CI again'],
- ['emoji-long-chars', '🚀'.repeat(70) + ' fix'],
- ['turkish-dotted', 'İstanbul office needs the plan reviewed today please'],
- ['accented-path', 'open über/café.py and tell me what it does please'],
- ['non-ascii-ext', 'look at 中文.py and 中文.PY and main.PY'],
- // paths and code
- ['path-home', 'update ~/notes/todo.md and ./src/app.ts and ../lib/util.js and /etc/hosts today'],
- ['path-colon-url', 'check https://example.com/docs and the file app.py today please'],
- ['path-ext-only', 'edit main.go then run build.sh and check Cargo.rs please'],
- ['path-many-effort', 'review a.py b.ts c.js d.md and e.json then verify the tests pass across all of them'],
- ['code-fence-only', '```\nx\n```'],
- ['code-fence-long', 'explain\n```py\nprint(1)\n```\nand\n```py\nprint(2)\n```\nin detail please now'],
- ]
- export default prompts.map(([name, prompt]) => ({ name, prompt }))
- // More cases for a prompt: [prompt name, options].
- export const extra = [
- ['complex-short', { parent: 'haiku' }],
- ['complex-short', { parent: 'sonnet', plan: true }],
- ['complex-code', { parent: 'sonnet' }],
- ['complex-code', { parent: 'fable' }],
- ['complex-code', { parent: 'opus', ctx: 700000 }],
- ['complex-code', { parent: 'sonnet', ctx: 650000 }],
- ['deep-design-paths', { parent: 'haiku' }],
- ['deep-design-paths', { parent: 'fable' }],
- ['deep-design-paths', { parent: 'sonnet', plan: true }],
- ['standard-fix', { parent: 'haiku' }],
- ['standard-fix', { parent: 'opus', ctx: 800000 }],
- ['standard-long', { parent: 'sonnet', transcript: 'synthetic-last' }],
- ['standard-long', { parent: 'opus', transcript: 'garbage' }],
- ['standard-long', { transcript: 'mythos' }],
- ['standard-long', { transcript: 'assistant-in-the-middle' }],
- ['fanout-numbered', { parent: 'haiku' }],
- ['fanout-verbs', { parent: 'opus', ctx: 900000 }],
- ['override-haiku', { parent: 'haiku' }],
- ['override-sonnet', { parent: 'opus' }],
- ['override-opus', { parent: 'haiku' }],
- ['override-fable', { parent: 'opus' }],
- ['override-fable', { parent: 'fable' }],
- ['ru-refactor', { parent: 'haiku' }],
- ['standard-fix', { config: { rules: [{ match: 'typo', model: 'haiku', reason: 'typos are cheap' }] } }],
- ['standard-fix', { parent: 'haiku', config: { rules: [{ match: 'TYPO', model: 'haiku' }] } }],
- ['deep-plan', { config: { rules: [{ match: 'rollout', model: 'sonnet', effort: 'low', confidence: 0.9 }] } }],
- ['complex-short', { config: { rules: [{ match: '^refactor', regex: true, model: 'fable', confidence: 0.7 }] } }],
- ['complex-short', { parent: 'sonnet', config: { rules: [{ match: '(', regex: true, model: 'haiku' }, { match: 'limiter', model: 'opus' }] } }],
- ['lookup-status', { parent: 'sonnet', config: { default_model: 'haiku' } }],
- ['lookup-list-prs', { config: { default_model: 'sonnet' } }],
- ['standard-fix', { config: { default_model: 'haiku' } }],
- ['standard-fix', { config: { disabled: true } }],
- ['standard-fix', { config: { rules: [{ match: 'readme', model: 'gpt-9', confidence: 0.6 }] } }],
- ['standard-fix', { config: { rules: [{ match: 'readme', model: 'haiku', confidence: 0.125 }] } }],
- ]
Ratings and comments
Sign in to rate this skill.
Sign inFree to use. If it’s helped, pass it on.