AutoCLI

AutoCLI is a terminal automation toolkit for developers and AI agents that turns websites, LLMs, developer platforms, editors, and utilities into one reusable CLI.

The core idea is simple:

sign in once
save the session or token locally
keep using the provider headlessly from the terminal
return clean --json output for scripts, agents, and orchestration

What makes AutoCLI especially useful is that it does not stop at API tokens. It works across cookies, saved browser sessions, user sessions, bot tokens, local tools, and public services, so the same CLI can drive GitHub, ChatGPT, Jira, Reddit, ffmpeg, DNS lookups, and more without switching tools.

Why It Matters

One command surface across 104 providers.
Shared browser login means less manual cookie exporting for cookie-backed platforms.
Sessions and tokens stay local, so follow-up commands are short and automation-friendly.
Category-based routing stays predictable as the tool grows: autocli llm ..., autocli social ..., autocli developer ..., autocli devops ....
Every provider is designed to be script-friendly, with strong --json support.
Provider capability metadata helps agents see auth type, stability, browser support, and read/write boundaries before they guess.
Shared result normalization adds stable JSON aliases like data.items, data.entity, and data.guidance.

Auto Browser Login

AutoCLI can keep a shared browser profile under its own control, let you sign in once, then reuse that browser state for later provider logins.

That means you can:

log into Google or another identity provider once
use Continue with Google, passkeys, or normal web sign-in flows
let later provider logins reuse that same saved browser profile
avoid re-exporting cookies every time for many cookie-backed providers

Typical flow:

autocli login --browser
autocli developer github login --browser
autocli social x login --browser
autocli llm qwen login --browser

After the provider session is saved, normal commands stay headless:

autocli developer github me --json
autocli social x post "Shipping from AutoCLI"
autocli llm qwen text "Summarize this changelog"

At a Glance

Item	Value
Package	`@vk007/autocli`
CLI command	`autocli`
Providers	`104`
Categories	`14`
npm install	`npm install -g @vk007/autocli`
bun install	`bun install -g @vk007/autocli`
Local setup	`bun install`

Get Started

Install globally with npm or Bun:

npm install -g @vk007/autocli
bun install -g @vk007/autocli

Validate the install right away:

autocli --version
autocli doctor
autocli doctor --fix

Set up the repo locally with Bun:

bun install
bun run build

Bootstrap the shared browser once if you want browser-assisted logins:

autocli login --browser

Typical commands:

autocli status
autocli llm chatgpt text "Write release notes for AutoCLI"
autocli developer github login --browser
autocli developer github me --json
autocli developer github capabilities --json
autocli devops cloudflare zones --json
autocli devops render services --json
autocli tools page-links https://example.com --json
autocli tools http github inspect --json

Every provider help page now includes:

a generated Quick Start block
a Support Profile with auth, discovery, mutation, browser, and async support
a Stability Guide so agents can tell whether a provider is stable, partial, or experimental

Why Use AutoCLI

Sign into real web apps once, then reuse the saved session from the terminal.
Use the same CLI for LLMs, socials, developer tools, devops platforms, editors, and public utilities.
Keep auth local to your machine instead of scattering cookies and tokens across one-off scripts.
Give agents and scripts a stable command model with consistent JSON output.
Reach protected web surfaces that are awkward to automate with plain APIs alone.

Command Model

AutoCLI is category-only. Provider commands never live at the root.

autocli llm ...
autocli editor ...
autocli finance ...
autocli maps ...
autocli movie ...
autocli news ...
autocli music ...
autocli social ...
autocli shopping ...
autocli developer ...
autocli devops ...
autocli bot ...
autocli tools ...

Examples:

autocli llm chatgpt text "Write release notes for AutoCLI"
autocli social x post "Shipping AutoCLI today"
autocli developer confluence search "release process"
autocli developer github me
autocli devops vercel projects
autocli bot telegrambot send 123456789 "Build finished"
autocli news top "AI"
autocli tools translate "hello world" --to hi
autocli tools timezone "Mumbai"
autocli tools oembed https://www.youtube.com/watch?v=dQw4w9WgXcQ
autocli tools http github request GET /settings/profile

Agent JSON Conventions

AutoCLI keeps provider-specific fields, but it also adds a few stable JSON aliases so agents can plan and transform results more reliably:

data.items for list-style results, even when the provider also returns keys like repos, projects, posts, or recommendations
data.entity for singular objects, even when the provider also returns keys like profile, page, movie, or project
data.meta.count and data.meta.listKey for quick list summaries
data.guidance.recommendedNextCommand and data.guidance.nextCommands for safer follow-up planning

Example:

autocli social reddit search "bun cli" --json
autocli movie tmdb title 27205 --json
autocli developer github capabilities --json

Stability Levels

stable: ready for routine automation and the default choice when you have options
partial: core flows work well, but some protected or edge routes may still need care
experimental: useful, but still changing quickly and best used with extra verification
unknown: not classified yet, so inspect with capabilities --json before leaning on it

To inspect a provider before acting:

autocli developer github capabilities --json
autocli social reddit capabilities --json
autocli devops railway capabilities --json

Category Overview

Category	Providers	Count	Needs	Why use it	Route
`llm`	ChatGPT, Claude, DeepSeek, Gemini, Grok, Mistral, Perplexity, Qwen, Z.ai	9	cookies	Browserless prompting, image flows, generation jobs	`autocli llm <provider> ...`
`editor`	Archive, Audio, Document, GIF, Image, PDF, Subtitle, Video	8	local tools	Media and file transformations from the terminal	`autocli editor <provider> ...`
`finance`	Crypto, Currency/Forex, Stocks	3	none	Market, forex, and crypto lookups	`autocli finance <provider> ...`
`data`	CSV, HTML, JSON, Markdown, Text, XML, YAML	7	none	Structured data cleanup, conversion, filtering, and extraction for agents	`autocli data <provider> ...`
`maps`	Geo, OpenStreetMap, OSRM	3	none	Geocoding, reverse lookup, routing, geometry helpers	`autocli maps <provider> ...`
`movie`	AniList, IMDb, JustWatch, Kitsu, Letterboxd, MyAnimeList, TMDb, TVMaze	8	none or cookies	Public title lookup, anime tracking, streaming availability, and community taste signals	`autocli movie <provider> ...`
`news`	News	1	none	Public headline discovery, source search, and feed aggregation	`autocli news ...`
`music`	Bandcamp, Deezer, SoundCloud, Spotify, YouTube Music	5	none or cookies	Public music discovery plus session-backed playback and library workflows	`autocli music <provider> ...`
`social`	Bluesky, Facebook, Instagram, LinkedIn, Mastodon, Pinterest, Reddit, Telegram, Threads, TikTok, WhatsApp, X, YouTube	13	none, cookies, or session	Public profile/thread lookup plus cookie-backed posting, Reddit discovery and write automation, federated discovery, MTProto messaging, and QR/session-backed chat control where supported	`autocli social <provider> ...`
`shopping`	Amazon, eBay, Etsy, Flipkart	4	none or cookies	Product discovery plus account/cart/order surfaces where supported	`autocli shopping <provider> ...`
`developer`	Confluence, GitHub, GitLab, Jira, Linear, Notion, Trello	7	cookies	Developer and workspace automation	`autocli developer <provider> ...`
`devops`	Cloudflare, DigitalOcean, Fly.io, Netlify, Railway, Render, Supabase, Vercel	8	api token	Infrastructure, deployment, DNS, platform, and backend automation	`autocli devops <provider> ...`
`bot`	Discord Bot, GitHub Bot, Slack Bot, Telegram Bot	4	bot token or app token	Notifications, chat ops, bot messaging	`autocli bot <provider> ...`
`tools`	Cheat, DNS, Favicon, Headers, HTTP Toolkit, IP, Markdown Fetch, Metadata, oEmbed, Page Links, QR, Redirect, Robots, RSS, Screenshot, Sitemap, SSL, Time, Timezone, Translate, Uptime, Weather, Web Search, Whois	24	none or cookies	Public utilities, session-aware request inspection, and zero-setup helpers	`autocli tools <provider> ...`

AutoCLI currently exposes 104 providers across 14 active command groups.

Access Modes

Needs	Meaning
`none`	Public or local functionality. No cookies, no token, no API key.
`local tools`	Uses binaries already installed on the machine, like `ffmpeg`, `ffprobe`, `qpdf`, or `yt-dlp`.
`cookies`	Import a browser session with `login --cookies ...` or let AutoCLI open a browser with `login --browser`, then reuse it headlessly.
`session`	Do one interactive login once, save the resulting user session locally, then reuse it headlessly.
`cookies + local token`	Cookie session plus a token the site keeps in localStorage or a similar client store.
`api token`	A personal or service token saved once with `login --token ...`.
`bot token`	A bot token saved once with `login --token ...`.
`browser later`	The current CLI route works for some surfaces, but more protected flows may later get an opt-in browser-backed mode.

Installation

Recommended Global Install

Use the published package as the primary supported install path:

npm install -g @vk007/autocli
bun install -g @vk007/autocli

After install, verify the command and your local environment:

autocli --version
autocli doctor
autocli doctor --fix
autocli status

autocli doctor checks the shared browser setup plus optional local tools such as ffmpeg, yt-dlp, qpdf, poppler, 7z, and macOS-native helpers when relevant.

On macOS, autocli doctor --fix can install all supported missing browser and local-tool dependencies automatically with Homebrew, then rerun the health check.

Local Development Setup

Set up the repo locally with Bun:

bun install

Build the Node-targeted bundle:

bun run build

Experimental Standalone Binary

You can also compile a standalone Bun binary:

bun run build:bin

This path is still experimental. The supported production install remains the npm or Bun global package above, because some runtime-heavy providers can behave differently in the compiled Bun binary.

Local Development Linking

Link autocli globally for local development:

bun run link:global

If your shell still says command not found, open a new shell or run hash -r.

Skill Doc Sync

Regenerate the provider-specific skill references and sync the installed Codex skill copy in one step:

bun run sync:skills

This refreshes the generated files under skills/autocli/references/providers and copies the repo skill into your local Codex skill directory (defaults to ~/.codex/skills/autocli unless CODEX_HOME is set).

npm publish now runs this automatically through prepublishOnly, so release builds regenerate and sync the skill docs before typecheck, tests, and build.

Open Source Project Files

If you plan to contribute, please do not commit live cookies, tokens, QR session state, or personal exports. AutoCLI should only store those locally on the contributor machine, never in the repository.

Quick Start

Check global status:

autocli status
autocli status --json
autocli doctor
autocli sessions

If you have not linked the CLI globally yet:

bun run dev status

Typical first-run flows:

autocli social x login --cookies ./x.cookies.json
autocli developer github login --browser
autocli llm chatgpt text "Summarize this changelog"
autocli developer github login --cookies ./github.cookies.json
autocli devops cloudflare login --token $CLOUDFLARE_API_TOKEN
autocli social telegram login --api-id 123456 --api-hash abcdef123456 --qr
autocli bot telegrambot login --token 123456:ABCDEF --name alerts-bot
autocli news top "AI"
autocli tools websearch search "bun commander zod"
autocli tools http github inspect

Best Example Workflows

Cookie-backed social posting

autocli social instagram login --cookies ./instagram.cookies.txt
autocli social instagram post ./photo.jpg --caption "Shipping from the terminal"
autocli social x login --cookies ./x.cookies.json
autocli social x post "Launching AutoCLI" --image ./launch.png

If you do not want to export cookies manually, many cookie-backed providers now also support:

autocli login --browser
autocli developer github login --browser
autocli social x login --browser
autocli llm qwen login --browser

autocli login --browser opens AutoCLI's shared browser profile so you can sign into Google or other identity providers once. Later provider logins reuse that same saved browser profile, and autocli <category> <provider> login --browser still skips opening the browser entirely when an already-saved active provider session is available.

LLM prompting and generation

autocli llm chatgpt text "Write release notes for AutoCLI"
autocli llm deepseek login --cookies ./deepseek.cookies.json --token <userToken>
autocli llm deepseek text "Explain retrieval-augmented generation"
autocli llm grok image "Minimal orange fox logo on white background"
autocli llm grok video "Minimal orange fox logo with subtle camera motion"

Developer and bot automation

autocli developer confluence search "deploy backend"
autocli developer github me
autocli developer gitlab projects "autocli" --limit 10
autocli developer jira projects
autocli developer linear issues --team ENG --limit 20
autocli developer trello boards
autocli devops netlify sites
autocli devops railway projects
autocli devops fly apps --org personal
autocli devops digitalocean apps
autocli bot telegrambot send 123456789 "Build finished"
autocli bot discordbot send 123456789012345678 "nightly deploy complete"

Session-backed messaging

autocli social telegram login --api-id 123456 --api-hash abcdef123456 --qr
autocli social telegram send me "Hello from AutoCLI"
autocli social reddit search "bun cli"
autocli social reddit post programming "Launching AutoCLI" "Now with Reddit support."
autocli social whatsapp login
autocli social whatsapp send 919876543210 "Ping from AutoCLI"

Public utilities

autocli news top "AI" --source google
autocli news search "typescript cli"
autocli news feed https://hnrss.org/frontpage --limit 5
autocli tools translate "hello world" --to hi
autocli tools websearch search "typescript cli bun"
autocli tools screenshot https://example.com --output-dir ./shots
autocli tools favicon openai.com
autocli tools page-links https://example.com --type external
autocli tools timezone "Mumbai"
autocli tools oembed https://www.youtube.com/watch?v=dQw4w9WgXcQ
autocli login --browser
autocli tools http github.com capture --browser-timeout 60
autocli tools http github.com capture --summary --group-by endpoint --browser-timeout 60
autocli tools uptime https://example.com --json
autocli tools rss https://hnrss.org/frontpage --limit 5

Music discovery and download

autocli music bandcamp search "radiohead"
autocli music bandcamp album https://radiohead.bandcamp.com/album/in-rainbows
autocli music soundcloud search "dandelions"
autocli music soundcloud user aviciiofficial
autocli music soundcloud playlist https://soundcloud.com/lofi-hip-hop-music/sets/lofi-lofi
autocli music soundcloud download "dandelions" --output-dir ./downloads

Local editing

autocli editor image resize ./photo.png --width 1200
autocli editor video split ./clip.mp4 --every 30
autocli editor audio loudness-report ./podcast.wav
autocli editor pdf watermark ./deck.pdf --text "Internal"
autocli editor subtitle burn ./video.mp4 --subtitle ./captions.srt

Sessions And Connections

Cookie sessions are stored under:

~/.autocli/sessions/<platform>/<account>.json

Token, bot, and saved session connections are stored under:

~/.autocli/connections/<platform>/<account>.json

AutoCLI supports importing:

Netscape cookies.txt
raw cookie strings
JSON cookie arrays
serialized tough-cookie jars

After the first login, later commands normally omit --account or --bot and AutoCLI uses the most recently saved connection for that provider.

Provider Matrix

The Stability column below matches the runtime metadata behind autocli <category> <provider> capabilities --json.

LLM

Provider	Stability	Needs	Best for	Notes
ChatGPT	`stable`	cookies	text prompts and image prompting	Authenticated web flow is supported; richer media generation is still evolving.
Claude	`partial`	cookies	Claude web-session access	Text prompting is the clearest supported path today; richer media flows still depend on private web-flow drift.
DeepSeek	`partial`	cookies + local token	browserless DeepSeek chat	Text chat works; image and video prompting are not mapped yet. Needs browser cookies and sometimes `userToken` from site storage.
Gemini	`stable`	cookies	Google Gemini prompting and media flows	Works from imported browser sessions; download helpers are wired for media jobs.
Grok	`partial`	cookies	text, image, and async video generation	Best current fit for job-style image/video workflows.
Mistral	`partial`	cookies	browserless Mistral chat	Cookie-backed text flow is implemented; image/video support is still limited.
Perplexity	`partial`	cookies	search-heavy prompting	Uses Perplexity’s live web flow for browserless text prompts; image/video support is still limited.
Qwen	`partial`	cookies	cookie-backed Qwen chat	Usually works directly from exported browser cookies.
Z.ai	`partial`	cookies	cookie-backed chat workflows	Session-backed provider for text-oriented usage today; media flows are still limited.

Editor

Provider	Stability	Needs	Best for	Notes
Archive Editor	`stable`	local tools	create, inspect, and extract archives	Good for zip/tar/gzip workflows.
Audio Editor	`stable`	local tools	trim, convert, normalize, denoise, report	Built around `ffmpeg` and related local tooling.
Document Editor	`stable`	local tools	document conversion, OCR, and text extraction	Useful for format conversion, OCR, and markdown export.
GIF Editor	`stable`	local tools	GIF create, optimize, and video conversion	Great for social clips and quick previews.
Image Editor	`stable`	local tools	image transformation pipelines	Includes upscale, background remove, watermark, compress, metadata strip, and more.
PDF Editor	`stable`	local tools	merge, split, watermark, reorder, secure	Also supports `to-images`, with better multi-page output when `pdftoppm` is installed.
Subtitle Editor	`stable`	local tools	shift, clean, merge, burn subtitles	Pairs well with YouTube and video publishing flows.
Video Editor	`stable`	local tools	split, stabilize, scene detect, overlay, transcode	Strong `ffmpeg`-driven video workflow surface.

Finance

Provider	Stability	Needs	Best for	Notes
Crypto	`stable`	none	crypto spot price lookup	No key required.
Currency	`stable`	none	currency conversion and forex math	Also exposed naturally as forex conversion.
Stocks	`stable`	none	stock quote lookups	Public market data surface.

Data

Provider	Stability	Needs	Best for	Notes
CSV	`stable`	none	inspect, filter, and convert CSV	Good for tabular exports and spreadsheet-style data.
HTML	`stable`	none	extract plain text or convert HTML to Markdown	Useful after scraping or page capture workflows.
JSON	`stable`	none	format, query, and merge JSON	Best glue layer for agent workflows across providers.
Markdown	`stable`	none	convert Markdown to HTML or text	Good for docs, release notes, and prompt shaping.
Text	`stable`	none	stats, replace, and dedupe lines	Useful for cleanup and normalization between steps.
XML	`stable`	none	format XML and convert it to JSON	Useful for feeds, sitemaps, and older structured formats.
YAML	`stable`	none	format YAML and convert it to JSON	Good for configs, manifests, and infra files.

Maps

Provider	Stability	Needs	Best for	Notes
Geo	`stable`	none	distance, midpoint, plus codes, elevation	Pure local/public helpers plus public elevation lookup.
OpenStreetMap	`stable`	none	search, reverse geocode, nearby lookup	Uses public OSM and Overpass services.
OSRM	`stable`	none	route, trip, table, nearest, match	Public routing and trip calculations.

Movie

Provider	Stability	Needs	Best for	Notes
AniList	`stable`	none	anime title lookup, trending, recommendations	Public anime metadata plus trending and recs.
IMDb	`stable`	none	movie and show lookup	Public search/title surface.
JustWatch	`stable`	none	where-to-watch checks	Streaming availability lookup.
Kitsu	`stable`	none	anime and manga discovery	Alternative anime metadata source.
Letterboxd	`stable`	none	film pages, diary feeds, and community taste	Public film pages plus profile and diary/RSS reading.
MyAnimeList	`stable`	none or cookies	public search plus your own saved list	Public lookups work without cookies; personal defaults can use cookies.
TMDb	`stable`	none	broad movie and TV title lookup	Uses TMDb's live public web catalog for search, title detail, popular titles, and recommendations.
TVMaze	`stable`	none	TV and episode-oriented title lookup	Strong TV-first public catalog with episode lists.

Music

Provider	Stability	Needs	Best for	Notes
Bandcamp	`stable`	none	public artist, album, and track lookup	Uses Bandcamp search plus readable album, track, and artist pages.
Deezer	`stable`	none	public track, album, artist, and playlist lookup	Uses Deezer's public entity API and a public search fallback where regional search is weak.
SoundCloud	`stable`	none	public track, playlist, and user discovery	Uses SoundCloud's public web client flow and can download tracks when a public stream is exposed.
Spotify	`stable`	cookies	library, playback, queue, playlists	Strongest playback-control surface in the repo today.
YouTube Music	`partial`	cookies	search and local playback control	Read commands can fall back to public browsing in some flows.

Social

Provider	Stability	Needs	Best for	Notes
Bluesky	`stable`	none	public profile, feed, and thread lookup	Uses Bluesky's public appview API, so it is cleaner and more stable than most social adapters.
Facebook	`partial`	cookies	session import and light account automation	Validation and safer read surfaces are strongest today; protected write flows are still conservative.
Instagram	`partial`	cookies	posting, downloads, stories, follows	One of the strongest cookie-backed social adapters in the repo.
LinkedIn	`partial`	cookies	posting and engagement	Text posting works best; richer media/write surfaces can drift and may need adapter refreshes over time.
Mastodon	`stable`	none	federated profile, posts, and thread lookup	Uses public instance APIs and respects the target instance when resolving accounts and statuses.
Pinterest	`stable`	none	public pin, profile, and board discovery	Good public-read discovery surface without cookies.
Reddit	`partial`	none or cookies	public post/thread discovery plus session-backed posting, comments, votes, and saves	Public reads use Reddit's JSON endpoints; writes can use a saved session or the shared browser profile with `--browser`.
Telegram	`partial`	session	QR, phone, or session-string login for account messaging	Uses MTProto with a saved user session, not browser cookies.
Threads	`partial`	none	public profile, post, and reply lookup	Uses the live Threads web surface through readable extraction; good for discovery, not write automation yet.
TikTok	`partial`	cookies	session handling and light public automation	Session validation and read flows are strongest today; signing for private web writes is still a known hard part.
WhatsApp	`partial`	session	QR or pairing-code login for account messaging	Uses a saved WhatsApp multi-device auth state with cached chats/messages for terminal workflows.
X	`partial`	cookies	posting, likes, profile/tweet lookup	Strong cookie-backed adapter with reliable agent-friendly output.
YouTube	`partial`	cookies	search, likes, comments, downloads	Upload/community posting are still separate future work.

Partial Support Notes

These providers are intentionally included, but their current browserless surface is narrower than their full website:

Claude, DeepSeek, Mistral, Perplexity, and Z.ai are strongest on text-oriented prompting today.
Facebook and TikTok are more reliable for session validation and lighter read flows than for protected web writes.
LinkedIn works best for text-first posting and engagement, but richer web flows can drift.
YouTube is strong for search, engagement, subscriptions, and downloads, while upload/community publishing is still separate future work.

Shopping

Provider	Stability	Needs	Best for	Notes
Amazon	`partial`	cookies	account, cart, add-to-cart, remove-from-cart, update-cart, search, product lookup	`add-to-cart`, `remove-from-cart`, `update-cart`, `cart`, `orders`, and `order` support browser-backed execution when the saved session alone is not enough.
eBay	`stable`	none	public listings, item lookup, seller discovery	Uses public readable page extraction plus the public autocomplete endpoint.
Etsy	`partial`	none	public listing and shop discovery	Direct Etsy fetches are anti-bot protected, so this adapter uses public site-search discovery today.
Flipkart	`stable`	cookies	account, wishlist, cart, add-to-cart, remove-from-cart, update-cart, orders	Uses the saved Flipkart session for cart actions. New adds use the authenticated cart endpoint; quantity updates and removals use the saved session in an invisible browser.

Developer

Provider	Stability	Needs	Best for	Notes
Confluence	`stable`	cookies	spaces, pages, search, comments, page creation	Uses an Atlassian web session to search and edit workspace documentation from the terminal.
GitHub	`stable`	cookies	repos, issues, pull requests, stars	Uses a saved GitHub web session for browserless developer automation.
GitLab	`stable`	cookies	projects, issues, merge requests	Uses a saved GitLab web session instead of personal access tokens.
Jira	`stable`	cookies	projects, issues, JQL search, issue creation	Saves a site-scoped Jira web session and reuses it for browserless workspace automation.
Linear	`partial`	cookies	issue management and comments	Uses the saved Linear web session for GraphQL issue workflows.
Notion	`partial`	cookies	search, pages, databases, comments	Uses the saved Notion web session instead of an official integration token.
Trello	`stable`	cookies	boards, lists, cards, card creation	Uses a saved Trello web session for browserless board and card workflows.

DevOps

Provider	Stability	Needs	Best for	Notes
Cloudflare	`stable`	api token	zones, DNS, and account-level infrastructure automation	Strong official API surface for DNS and zone operations.
DigitalOcean	`stable`	api token	App Platform, deployments, and domains	Uses the official App Platform and account APIs.
Fly.io	`partial`	api token	apps, machines, volumes, and certificates	Some tokens may need an explicit `--org` slug for the cleanest results.
Netlify	`stable`	api token	sites, deploys, DNS, and team account automation	Good fit for deployment and static hosting workflows.
Railway	`partial`	api token	projects, services, and environment-oriented platform workflows	Uses Railway’s GraphQL surface, so deeper actions may still expand over time.
Render	`stable`	api token	services, projects, and env-group automation	Strong official REST API for deployment and service inspection.
Supabase	`stable`	api token	organizations, projects, and functions	Good backend/platform automation surface.
Vercel	`stable`	api token	teams, projects, and deployments	Strong deployment-centric API coverage.

Bot

Provider	Stability	Needs	Best for	Notes
Discord Bot	`stable`	bot token	guild/channel messaging automation	Good for notifications and chat ops.
GitHub Bot	`stable`	api token	GitHub app or bot-token actions	Same GitHub surface, but bot-style auth.
Slack Bot	`stable`	bot token	channel history, send, edit, file upload	Strong for workspace notification flows.
Telegram Bot	`stable`	bot token	send, edit, media, updates	Good default bot adapter for lightweight notifications.

Tools

Provider	Stability	Needs	Best for	Notes
Cheat	`stable`	none	shell and language cheat sheets	Fast terminal help.
DNS	`stable`	none	DNS record lookups	Good for quick ops checks.
Favicon	`stable`	none	site icon discovery and verification	Resolves declared icon tags, falls back to `/favicon.ico`, and verifies candidates.
Headers	`stable`	none	inspect raw HTTP response headers	Useful for cache, CDN, and server debugging.
HTTP Toolkit	`stable`	none or cookies	inspect saved sessions, attach to the shared browser, and replay authenticated requests	Best for session-aware debugging on sites where plain `curl` is not enough.
IP	`stable`	none	public IP and network details	Fast no-auth network lookup.
Markdown Fetch	`stable`	none	turn pages into markdown-like text	Useful for scraping readable content.
Metadata	`stable`	none	webpage title and social tags	Extracts title, description, canonical, favicon, Open Graph, and Twitter tags.
oEmbed	`stable`	none	embeddable media/page metadata from URLs	Uses page-discovered oEmbed endpoints first, then falls back to a public resolver.
Page Links	`stable`	none	internal/external link extraction from webpages	Useful for crawls, site audits, and agent discovery.
QR	`stable`	none	QR generation	Can save or print a public image URL.
Redirect	`stable`	none	redirect-chain tracing	Shows each HTTP hop and final destination.
Robots	`stable`	none	`robots.txt` inspection	Useful for site crawling checks.
RSS	`stable`	none	feed inspection	Reads RSS/Atom without setup.
Screenshot	`stable`	none	URL-to-image captures	Public no-key render service.
Sitemap	`stable`	none	sitemap discovery and listing	Good for SEO/crawl inspection.
SSL	`stable`	none	TLS certificate inspection	Shows certificate, issuer, SANs, protocol, and expiry.
Time	`stable`	none	timezone and current time lookup	Public time APIs.
Timezone	`stable`	none	resolve timezone from a place, coordinates, or IANA zone	Useful for scheduling, agent routing, and regional workflows.
Translate	`stable`	none	quick translation	Uses a public translation endpoint.
Uptime	`stable`	none	latency and HTTP health checks	Lightweight monitoring helper.
Weather	`stable`	none	weather lookup	No account required.
Web Search	`stable`	none	multi-engine search	Supports summaries and engine selection.
Whois	`stable`	none	domain registration details	Useful for domain inspection.

News

Provider	Stability	Needs	Best for	Notes
News	`stable`	none	headline and feed aggregation	Pulls from public no-key sources like Google News, GDELT, Hacker News, Reddit, and raw RSS feeds.

Agent-Friendly Output

Every command supports --json.

Success shape:

{
  "ok": true,
  "platform": "x",
  "account": "personal",
  "action": "post",
  "message": "X post created for personal.",
  "id": "1234567890",
  "url": "https://x.com/user/status/1234567890"
}

Error shape:

{
  "ok": false,
  "error": {
    "code": "SESSION_EXPIRED",
    "message": "X returned a logged-out page. Re-import cookies.txt."
  }
}

This makes AutoCLI friendly for:

shell scripts
CI jobs
multi-step agents
external orchestrators

Session Refresh

AutoCLI includes a refresh layer in src/utils/autorefresh.ts.

Instagram, X, and YouTube can use lightweight authenticated keepalive checks before normal actions.
Rotated cookies are persisted back into the saved session file when the platform returns them.
Some platforms still do not support a durable cookie-only refresh path, so expiration handling remains provider-specific.

This is the most practical browserless approach for copied web sessions, but it is not a universal guarantee for every website.

For cookie-backed providers that support interactive capture, you can also use login --browser to open a real browser, complete the sign-in flow manually, and let AutoCLI save the session automatically.

Project Structure

.
├── README.md
├── package.json
├── tsconfig.json
└── src
    ├── __tests__
    ├── core
    │   ├── auth
    │   └── runtime
    ├── commands
    ├── platforms
    │   ├── bot
    │   ├── developer
    │   ├── editor
    │   ├── finance
    │   ├── llm
    │   ├── maps
    │   ├── movie
    │   ├── music
    │   ├── shopping
    │   ├── social
    │   ├── tools
    │   ├── config.ts
    │   └── index.ts
    ├── utils
    ├── config.ts
    ├── errors.ts
    ├── index.ts
    └── logger.ts

Development

Run the local CLI:

bun run dev --help
bun run dev social x login --cookies ./cookie.json

Watch mode:

bun run dev:watch --help

Typecheck:

bun run typecheck

Run tests:

bun test

Notes On Reliability

Cookie-backed private web flows can drift as providers change internal endpoints.
Token-backed developer and bot providers are usually the most stable long-term.
Local editor and public utility providers are the least fragile because they do not depend on private web sessions.
The category model is intentionally strict so provider names do not collide as AutoCLI grows.

Name		Name	Last commit message	Last commit date
Latest commit History 212 Commits
.github		.github
scripts		scripts
skills/autocli		skills/autocli
src		src
.gitignore		.gitignore
CONTRIBUTING.md		CONTRIBUTING.md
LICENSE		LICENSE
README.md		README.md
SECURITY.md		SECURITY.md
bun.lock		bun.lock
package.json		package.json
tsconfig.json		tsconfig.json

Folders and files

Latest commit

History

Repository files navigation

AutoCLI

Why It Matters

Auto Browser Login

At a Glance

Get Started

Why Use AutoCLI

Command Model

Agent JSON Conventions

Stability Levels

Category Overview

Access Modes

Installation

Recommended Global Install

Local Development Setup

Experimental Standalone Binary

Local Development Linking

Skill Doc Sync

Open Source Project Files

Quick Start

Best Example Workflows

Cookie-backed social posting

LLM prompting and generation

Developer and bot automation

Session-backed messaging

Public utilities

Music discovery and download

Local editing

Sessions And Connections

Provider Matrix

LLM

Editor

Finance

Data

Maps

Movie

Music

Social

Partial Support Notes

Shopping

Developer

DevOps

Bot

Tools

News

Agent-Friendly Output

Session Refresh

Project Structure

Development

Notes On Reliability

Star History

About

Topics

Resources

License

Contributing

Security policy

Uh oh!

Stars

Watchers

Forks

Releases

Packages 0

Uh oh!

Contributors

Uh oh!

Languages

Packages