Manage headless backend services including databases, authentication, file storage, workflows, and payments through an AI-ready interface.
Also known as anythink-cloud/anythink-cli, anythink-cli
- Stars
- 6
- Tools
- 15
- Downloads, 30 days
- —
- Latest version
- 0.2.37
- Last commit
- Nov 5, 2026
- License
- MIT
What is Anythink MCP?
Anythink manages headless backend services including databases, authentication, file storage, workflows, and payments through an AI-ready interface. It exposes 15 tools, such as signup, login, login_direct.
Summary generated from the project's README
About Anythink MCP server
Anythink provides a headless backend platform designed for developers and founders to manage complex infrastructure through a single interface. The MCP server allows AI clients to interact directly with backend services including databases, authentication, file storage, workflows, and payments. Users can manage their entire backend lifecycle without leaving their development environment by using the provided tools to control projects and entities.
Setup involves installing the anythink-mcp binary via Homebrew on macOS or Linux, or by downloading the platform-specific binaries for Apple Silicon, Intel, or various Linux architectures. Once installed, users must sign up for an account and create a billing account to host projects. The system uses a local configuration file at ~/.anythink/config.json to store credentials and project profiles, allowing for easy switching between different isolated backend instances using the projects use command.
What you can do with it
- 1Manage database schemasCreate and configure database tables and their specific columns using the entities and fields commands.
- 2Configure project infrastructureSet up new isolated backend instances in specific deployment regions with chosen subscription plans.
Written from the project's README. Every use case is checked against a quote from it.
How to install Anythink MCP
Formats verified 10 Nov 2026Add to CursorAdd to GooseInstall Anythink MCP in Claude Code
$ claude mcp add --transport http anythink-cli https://mcp.anythink.cloud/mcpInstall Anythink MCP in Cursor
~/.cursor/mcp.json
{
"mcpServers": {
"anythink-cli": {
"url": "https://mcp.anythink.cloud/mcp"
}
}
}Install Anythink MCP in VS Code
.vscode/mcp.json
{
"servers": {
"anythink-cli": {
"type": "http",
"url": "https://mcp.anythink.cloud/mcp"
}
}
}Install Anythink MCP in VS Code CLI
$ code --add-mcp '{"name":"anythink-cli","type":"http","url":"https://mcp.anythink.cloud/mcp"}'Install Anythink MCP in Codex
$ codex mcp add anythink-cli --url https://mcp.anythink.cloud/mcpInstall Anythink MCP in Codex config.toml
~/.codex/config.toml
[mcp_servers.anythink-cli] url = "https://mcp.anythink.cloud/mcp"
Install Anythink MCP in Gemini CLI
$ gemini mcp add -s user --transport http anythink-cli https://mcp.anythink.cloud/mcpInstall Anythink MCP in Gemini settings.json
~/.gemini/settings.json
{
"mcpServers": {
"anythink-cli": {
"url": "https://mcp.anythink.cloud/mcp",
"type": "http"
}
}
}Anythink MCP tools 15
Detected from the READMEsignuploginlogin_directlogoutconfig_showconfig_useconfig_removeaccounts_listaccounts_createaccounts_useprojects_listprojects_create
| signup | — | |
| login | — | |
| login_direct | — | |
| logout | — | |
| config_show | — | |
| config_use | — | |
| config_remove | — | |
| accounts_list | — | |
| accounts_create | — | |
| accounts_use | — | |
| projects_list | — | |
| projects_create | — |
Show all 15 toolsShow fewer tools
projects_useprojects_deletecli
| projects_use | — | |
| projects_delete | — | |
| cli | — |
Automated checks
Ran Nov 12, 2026Full report →Passed our automated checks. That is not a security audit: read the code and give the server the narrowest access it needs. How checks work
- Source repository linked and reachablePass
- Open source license (OSI)Pass
- Commit or release in the last yearPass
- Package name not close to a popular packageNot checked
- Remote endpoint uses HTTPSPass
- Destructive tools are labelledPass
- No hidden instructions to the modelPass
- No critical advisory in the package or its dependenciesNot checked
README
View on GitHubAnythink CLI
The official command-line interface for Anythink — the headless backend platform for developers and founders. Manage your projects, entities, data, workflows, users, files, and payments without leaving the terminal.
░███ ░██ ░██ ░██ ░██
░██░██ ░██ ░██ ░██
░██ ░██ ░████████ ░██ ░██ ░████████ ░████████ ░██░████████ ░██ ░██
░█████████ ░██ ░██ ░██ ░██ ░██ ░██ ░██ ░██░██ ░██ ░██ ░██
░██ ░██ ░██ ░██ ░██ ░██ ░██ ░██ ░██ ░██░██ ░██ ░███████
░██ ░██ ░██ ░██ ░██ ░███ ░██ ░██ ░██ ░██░██ ░██ ░██ ░██
░██ ░██ ░██ ░██ ░█████░██ ░████ ░██ ░██ ░██░██ ░██ ░██ ░██
░██
░███████
Contents
Installation
Both the anythink CLI and the anythink-mcp server are distributed together. The quickest way to get both is Homebrew.
Homebrew (macOS / Linux) — recommended
brew install anythink-cloud/tap/anythink
This installs both commands onto your PATH:
anythink— the CLIanythink-mcp— the MCP server (see MCP server)
Or tap first, then install:
brew tap anythink-cloud/tap
brew install anythink
Update to the latest release:
brew upgrade anythink
Uninstall:
brew uninstall anythink
macOS / Linux — download binary
Grab the latest release for your platform from the Releases page:
| Platform | Binary |
|---|---|
| macOS (Apple Silicon) | anythink-osx-arm64 |
| macOS (Intel) | anythink-osx-x64 |
| Linux (x86_64) | anythink-linux-x64 |
| Linux (ARM64) | anythink-linux-arm64 |
# Example — macOS Apple Silicon
curl -Lo anythink https://github.com/anythink-cloud/anythink-cli/releases/latest/download/anythink-osx-arm64
chmod +x anythink
sudo mv anythink /usr/local/bin/
The MCP server ships as a matching anythink-mcp-<platform> binary on the same release — download and install it the same way (e.g. anythink-mcp-osx-arm64).
Verify the download against checksums.txt in the release assets:
sha256sum -c checksums.txt --ignore-missing
.NET global tool
If you have the .NET 8 SDK installed:
dotnet tool install --global anythink-cli
Build from source
git clone https://github.com/anythink-cloud/anythink-cli
cd anythink-cli
dotnet build
dotnet run --project src/AnythinkCli.csproj -- --help
Getting started
# 1. Create an account (or log in if you already have one)
anythink signup
# 2. Create a billing account
anythink accounts create --name "My Company"
# 3. Create a project
anythink projects create "My App" --region lon1
# 4. Connect to the project
anythink projects use <project-id>
# 5. Start building
anythink entities list
anythink workflows list
Your credentials and project profiles are stored in ~/.anythink/config.json. You can manage multiple projects by running projects use to switch between them.
Command reference
signup / login / logout
anythink signup Create a new Anythink account
anythink login Log in to the platform
anythink logout Remove saved credentials for a project profile
signup and login are interactive — they prompt for email and password and walk you through connecting to a billing account and project on first run.
accounts
Manage billing accounts. A billing account is the container for one or more projects and holds your subscription and payment details.
anythink accounts list List your billing accounts
anythink accounts create Create a new billing account
anythink accounts use <id> Set the active billing account
Examples
anythink accounts create --name "Acme Ltd" --email billing@acme.com
anythink accounts use a1b2c3d4
projects
Create and manage Anythink projects. Each project is an isolated backend instance with its own database, auth, files, and workflows.
anythink projects list List projects in the active billing account
anythink projects create <name> Create a new project
anythink projects use <id> Connect to a project (sets it as the active profile)
anythink projects delete <id> Delete a project
Options — projects create
| Flag | Description |
|---|---|
--region <id> |
Deployment region (e.g. lon1) |
--plan <id> |
Plan ID (see anythink plans) |
Examples
anythink projects create "My App" --region lon1
anythink projects use a1b2c3d4
anythink projects delete a1b2c3d4 --yes
config
View and manage saved CLI profiles.
anythink config show List all profiles and platform settings
anythink config use <profile> Set the active project profile
anythink config remove <profile> Remove a profile
Profiles are named by project alias or ID and stored in ~/.anythink/config.json.
entities
Manage entities (database tables) in the active project.
anythink entities list List all entities
anythink entities get <name> Get entity details and fields
anythink entities create <name> Create a new entity
anythink entities update <name> Update entity settings
anythink entities delete <name> Delete an entity and all its data
Options — entities create
| Flag | Description |
|---|---|
--rls |
Enable row-level security |
--public |
Make the entity publicly readable |
--lock |
Lock new records (prevent direct creation) |
--junction |
Mark as a junction (many-to-many) table |
Examples
anythink entities create orders --rls
anythink entities get customers
anythink entities delete temp_data --yes
fields
Manage fields on an entity. Fields map directly to database columns.
anythink fields list <entity> List fields on an entity
anythink fields add <entity> <name> Add a field
anythink fields delete <entity> <id> Delete a field
Options — fields add
| Flag | Description |
|---|---|
--type <type> |
Field type: varchar, text, int, float, bool, datetime, json, uuid |
--required |
Mark the field as required |
--unique |
Enforce a unique constraint |
--indexed |
Add a database index |
--default <value> |
Default value |
Examples
anythink fields list customers
anythink fields add customers email --type varchar --unique --required
anythink fields add products price --type float --required
anythink fields delete customers 1234 --yes
data
CRUD operations on entity records.
anythink data list <entity> List records
anythink data get <entity> <id> Get a single record by ID
anythink data create <entity> Create a new record
anythink data update <entity> <id> Update a record
anythink data delete <entity> <id> Delete a record
anythink data export <entity> <file> Export records to CSV, JSON or JSONL
anythink data import <entity> <file> Create records from CSV, JSON or JSONL
Options — data list
| Flag | Description |
|---|---|
--limit <n> |
Records per page (default: 20) |
--page <n> |
Page number (default: 1) |
--filter <expr> |
Field filters — JSON ({"status":"draft","total":{"gte":10}}) or field=value pairs (status=draft&total=GTE:10). See below. |
--json |
Output raw JSON instead of table |
--all |
Stream all pages as sequential JSON objects (requires --json, constant memory) |
Filter operators (JSON form) — {"field": value} matches exactly, {"field": [a, b]} matches any, {"field": null} matches empty. For anything else use an operator object, e.g. {"price": {"gte": 10, "lt": 100}}:
| Operator | Meaning | field=value form |
|---|---|---|
eq / ne |
Equal / not equal | x=v / x=!v |
gt gte lt lte |
Comparisons | x=GT:v … |
contains / not_contains |
Case-insensitive substring | x=C:v / x=NC:v |
starts_with / ends_with |
Case-insensitive prefix/suffix | x=SW:v / x=EW:v |
in |
Any of a list | x=IN:a,b |
null: true / null: false |
Empty / not empty | x=NULL: / x=NNULL: |
exists: true / exists: false |
Has / lacks related records | x=EXISTS: / x=NOT_EXISTS: |
Options — data create / data update
| Flag | Description |
|---|---|
--data <json> |
JSON object of field values |
Examples
anythink data list blog_posts --limit 10
anythink data get blog_posts 42
anythink data create blog_posts --data '{"title":"Hello World","status":"draft"}'
anythink data update blog_posts 42 --data '{"status":"approved"}'
anythink data delete blog_posts 42 --yes
Export and import — data export streams records page by page to a file; data import creates one record per row. Always run an import with --dry-run first: it validates every row and sends nothing.
anythink data export customers customers.csv
anythink data export orders orders.jsonl --filter 'status=paid' --fields id,total
anythink data import customers customers.csv --dry-run
anythink data import customers customers.csv --errors rejected.csv --yes
Flag (export) |
Description |
|---|---|
--format <fmt> |
csv, json or jsonl. Default: from the extension (.csv, .json, .jsonl, .ndjson). - as the file writes to stdout (needs --format) |
--filter <expr> |
Same syntax as data list --filter |
--fields a,b |
Columns to export, in order. CSV defaults to the entity's fields |
--page-size <n> |
Records per request (default 200, max 1000) |
--force |
Overwrite an existing file. Without it an existing file is never touched. The file is written to a temporary name and renamed on success, so a failed export never leaves a partial file |
--no-formula-guard |
CSV only: see below |
Flag (import) |
Description |
|---|---|
--format <fmt> |
csv, json (an array of objects) or jsonl. Default: from the extension, else the first character ([ or {) |
--dry-run |
Validate and report; nothing is sent |
--concurrency <n> |
Parallel creates (default 4, max 16) |
--errors <file> |
Write skipped and failed rows, with an _error column or property, so they can be fixed and re-imported |
--ignore-unknown |
Drop columns that aren't fields on the entity instead of rejecting the row |
--yes |
Skip the confirmation prompt. Required when not running interactively (CI, scripts, agents), otherwise the import refuses before writing anything |
How values are handled:
- CSV is RFC 4180: cells containing a comma, quote or line break are quoted, quotes are doubled, and quoted cells can span lines. Objects, arrays, JSON fields and relations are written as compact JSON text.
- Formula guard (export): CSV text starting with
=,+,-,@, tab or carriage return is prefixed with'so spreadsheets don't run it as a formula. Numbers are never changed. This includes phone numbers such as+44 20 7946 0000and any text with a leading+or-. The guard is one-way: importing the file keeps the'unless the file was exported with--no-formula-guard. - Types follow the entity's fields. Text keeps its exact characters (
00123stays00123); integers and decimals are parsed with.as the decimal separator (1,5is rejected, never guessed); booleans accepttrue/false,1/0andyes/no; dates must be ISO 8601 (2025-03-31,2025-03-31T09:30:00Z; no offset means UTC); JSON fields must be valid JSON. - An empty CSV cell is sent as
null. In JSON and JSONL input an empty string stays an empty string. - Bad rows are skipped, never sent. They are reported with a row number (CSV: record number, so the header is row 1; JSONL: line number; JSON: position in the array) and counted as skipped. Rows the server rejects are counted as failed. The exit code is non-zero if any row was skipped or failed.
- Retries:
400,409and422are never retried, so a create is not repeated. See Uncertain failures below for what is retried. - System columns are ignored on import:
id,created_at,updated_at,tenant_idandlockedare never sent, and the import says which it ignored. Importing an export back into the same project therefore creates duplicates with new ids and timestamps; it does not update the originals. - One-to-many collections and secret fields are skipped on import (the import lists them), and one-to-many columns are left out of an export unless you name them in
--fields. Many-to-many, file and user columns are copied as ids, which only mean something inside the same project; the import warns once per column. - Empty boolean cells are sent as
nullrather thanfalse. - Uncertain failures: only
429/503responses and requests that never reached the server (DNS, connection or TLS failures) are retried. If a request times out or the connection drops after sending, the row is reported asUNCERTAIN: may have been created, in the errors file too; check the project before re-importing it. Rate-limit waits pause every worker together and are retried for over a minute before giving up. - Ctrl+C stops reading the file, finishes the rows already queued and prints a partial summary (exit code 1). An interrupted export removes its temporary file.
data importanddata exportread and write local files, so they are not available through the hosted (HTTP) MCP server.
Not supported yet: updating or upserting existing records by key, looking up relations by a natural key (use the related record's id), and file uploads.
search
Full-text search across your entities, plus index lifecycle management.
anythink search query <text> Run a search
anythink search similar <entity> <id> Find similar documents
anythink search rehydrate [<entity>] Rebuild the search index (admin)
anythink search purge [<entity>] Wipe the search index (admin)
anythink search audit <entity> Compare configured public-searchable fields
with what public search actually returns
Options — search query
| Flag | Description |
|---|---|
--entities <list> |
Comma-separated entity names. Default: all indexed entities. |
--filter <expr> |
Filter expression, e.g. "status=published AND category=news". Supports _geoRadius. |
--sort <list> |
Comma-separated sort fields, e.g. "created_at:desc,id:asc". |
--facet <fields> |
Comma-separated fields to compute facet counts on. |
--highlight |
Highlight matched terms in results. |
--page N |
Page number (default: 1). |
--limit N |
Results per page (1-100, default: 20). |
--public |
Use the unauthenticated /search/public endpoint (only public-marked fields). |
--json |
Print the raw response JSON. |
Index lifecycle
rehydrate and purge are admin operations on the search index:
search rehydrate— rebuilds the index from the database (no data loss; just resyncs)search purge— deletes the index (runrehydrateafter to repopulate)
Both confirm by default; pass -y / --yes to skip the prompt for automation.
search audit — public-search data leak check
Compares what the entity's schema says should be public-searchable (fields with publicly_searchable=true and the entity's own is_public=true) against what /search/public actually returns. Any field appearing in public results that isn't on the allowlist is reported as a leak.
Exits with code 1 if a leak is detected — useful for CI/CD.
Examples
# Browse everything
anythink search query "*"
# Filtered search with sorting
anythink search query "anythink" --filter "status=published" --sort "created_at:desc"
# Compare what public visitors see vs what's in the database
anythink search audit posts
anythink search audit users --query "alice" --sample 10
# Reindex after a schema change
anythink search rehydrate posts
anythink search rehydrate --yes # everything (admin)
# Geo search (radius in metres)
anythink search query "*" --filter "_geoRadius(51.5074,-0.1278,5000)"
workflows
Manage automation workflows. A workflow can have several triggers: a cron schedule, an entity event, an API route, or a manual run.
anythink workflows list List all workflows (--json for a compact summary)
anythink workflows get <id> Get workflow details and steps (--json for the full definition)
anythink workflows create <name> Create a new workflow
anythink workflows update <id> Rename a workflow or change its description
anythink workflows enable <id> Enable a workflow
anythink workflows disable <id> Disable a workflow
anythink workflows trigger <id> Manually trigger a workflow
anythink workflows delete <id> Delete a workflow
Trigger types — workflows create --trigger <type> (not case-sensitive; default Manual)
| Type | Fires | Required flag | Other flags |
|---|---|---|---|
Manual |
When run by hand, on an entity | --entity <name> |
|
Event |
When an event happens, such as a record being created | --entity <name> for the Entity... events |
--event <event>, --filter <json> |
Timed |
On a cron schedule | --cron <expr> |
|
Api |
When its API route is called | --api-route <route> |
--event (not case-sensitive) is EntityCreated (default), EntityUpdated, EntityDeleted, UserRegistered, UserInvited, SubscriptionCreated, SubscriptionActivated, SubscriptionExpired, PaymentCreated, PaymentSucceeded, PaymentFailed, PaymentMethodCaptured, PaymentMethodCaptureFailed, PaymentMethodRemoved or PushActionTaken; only the three Entity... events need --entity. A trigger missing its required flag is rejected before anything is sent, and a flag that doesn't apply to the chosen type is ignored with a warning.
Other options — workflows create
| Flag | Description |
|---|---|
--description <text> |
Workflow description |
--enabled |
Enable the workflow straight away |
--filter-file <path> |
Read the Event filter JSON from a file |
Examples
anythink workflows create daily-sync --trigger Timed --cron "0 6 * * *"
anythink workflows create on-post --trigger Event --entity blog_posts --event EntityUpdated
anythink workflows create import-hook --trigger Api --api-route hooks/import
anythink workflows create review-posts --entity blog_posts
anythink workflows trigger 76
anythink workflows disable 83
Anythink MCP questions
How do I add Anythink MCP to Claude Code?
Run claude mcp add --transport http anythink-cli https://mcp.anythink.cloud/mcp in your terminal. Add --scope user to make it available in every project.
How do I add Anythink MCP to Cursor?
Open ~/.cursor/mcp.json and add the server under mcpServers, using the Cursor config shown in the install section above.
What tools does Anythink MCP provide?
It exposes 15 tools, including signup, login, login_direct, logout, config_show.
What can I manage with Anythink?
You can manage projects, entities, data, workflows, users, files, and payments.
How do I install the Anythink CLI and MCP server?
The easiest way is using Homebrew on macOS or Linux, which installs both the CLI and the MCP server.
Can I create a new project with a specific deployment region?
Yes, you can specify a deployment region using the --region flag when creating a project.
How are my credentials stored?
Your credentials and project profiles are stored locally in a JSON file at ~/.anythink/config.json.