Connect your data sources

Anything CLI queries data where it already lives — Postgres, MySQL, MongoDB, Salesforce, CSV files, and more. You do not paste connection strings into playbooks or chat. Instead, you register named sources in profiles/local.yaml and keep secrets in .env.

Connection flow
1. Put credentials in .env (gitignored)
2. Register each source in profiles/local.yaml with an adapter type
3. Restart the stack (anythinggraph start or ./start-all.sh)
4. Map playbook entities to sources in bindings via source_id

Profiles are not written via MCP. Edit profiles/local.yaml by hand (or in your deployment repo). MCP admin tools can list and introspect sources after they are configured — they do not create profile entries for you.

Where is `local.yaml`?

The profile file is always named local.yaml and lives under a profiles/ directory in your Anything CLI workspace.

Install method	Profile path
npm CLI (`anythinggraph onboard`)	`~/.anythinggraph/source/profiles/local.yaml`
Git clone (contributors)	`./profiles/local.yaml` in the repo root
Custom path	Set `AG_PROFILE_PATH` to your YAML file

Credentials live next to the workspace in .env — for npm installs: ~/.anythinggraph/source/.env. Copy from .env.example if you have not already.

# npm CLI install
cp ~/.anythinggraph/source/.env.example ~/.anythinggraph/source/.env
# edit AG_SQL_DSN, AG_MYSQL_DSN, etc.

# git clone
cp .env.example .env

Profile structure

A profile is a YAML file with a top-level sources: map. Each key is a source id you choose (for example warehouse_pg, mysql_main). Bindings reference that key as source_id.

Shipped example (from the repo)

sources:
  warehouse_pg:
    adapter: sql
    dsn: env:AG_SQL_DSN
  salesforce_main:
    adapter: soql
    instance_url: env:AG_SF_INSTANCE_URL
    auth: env:AG_SF_ACCESS_TOKEN
  payroll_csv:
    adapter: csv
    file_path: env:AG_PAYROLL_CSV_PATH
  mongo_main:
    adapter: mongodb
    dsn: env:AG_MONGODB_DSN
    database: env:AG_MONGODB_DATABASE

Field	Purpose
`sources.<key>`	Named source — becomes `source_id` in bindings
`adapter`	Which connector to use: `sql`, `mysql`, `mongodb`, `csv`, `soql`, `rest`, …
`dsn`	Database connection string (SQL, MySQL, MongoDB)
`file_path`	Path to a CSV file (CSV adapter)
`instance_url` + `auth`	Salesforce org URL and access token (SOQL adapter)
`database`	MongoDB database name (optional; can also pass at introspect time)
`env:AG_*`	Prefix meaning “read this value from an environment variable at startup”

How profiles work

Secrets stay in .env — connection strings, tokens, and file paths. Never put them in playbooks, bindings, or git.
profiles/local.yaml names each source — adapter type + env:… references only.
Bindings point at a profile key — source_id: warehouse_pg tells the runtime which credentials and adapter to use for that mapping file.
One playbook, many sources — you often have one binding YAML per source (for example my-playbook.postgres.yaml and my-playbook.csv.yaml), each with its own source_id.

Example binding reference — after connecting Postgres, a binding might start with: source_id: warehouse_pg. See Bindings and the playbooks guide for entity mapping.

Quick checklist (every source)

Add the credential to .env
Add or edit an entry under sources: in profiles/local.yaml
Restart: anythinggraph start or ./start-all.sh
Confirm with MCP list_sources (admin token) or anythinggraph doctor
Author a binding with source_id matching your profile key

MySQL

1. Set credentials in `.env`

AG_MYSQL_DSN=mysql://user:pass@localhost:3306/yourdb

2. Register in `profiles/local.yaml`

sources:
  mysql_main:
    adapter: mysql
    dsn: env:AG_MYSQL_DSN

3. Use in a binding

# bindings/your-playbook.mysql.yaml
source_id: mysql_main

entities:
  crm_user:
    from: users
    id: user_id
    fields: [full_name]

from is the table name; id is the primary key column. SQL is compiled for MySQL dialect automatically. Introspect via MCP: introspect_source with source_id: mysql_main and optional schema_name for the database schema.

Salesforce

1. Set credentials in `.env`

AG_SF_INSTANCE_URL=https://your-org.my.salesforce.com
AG_SF_ACCESS_TOKEN=your_access_token

2. Register in `profiles/local.yaml`

sources:
  salesforce_main:
    adapter: soql
    instance_url: env:AG_SF_INSTANCE_URL
    auth: env:AG_SF_ACCESS_TOKEN

3. Use in a binding

# bindings/your-playbook.salesforce.yaml
source_id: salesforce_main

entities:
  crm_user:
    from: User
    id: Id
    fields:
      full_name: Name

from is the Salesforce object API name (User, Account, Lead, …). Field names use Salesforce API names. Introspect with introspect_source and optional schema_name=Account,Contact to limit described objects.

CSV files

1. Set file path in `.env`

AG_PAYROLL_CSV_PATH=./data/payroll.csv

Use an absolute path or a path relative to the workspace root when the service starts.

2. Register in `profiles/local.yaml`

sources:
  payroll_csv:
    adapter: csv
    file_path: env:AG_PAYROLL_CSV_PATH

3. Use in a binding

# bindings/your-playbook.csv.yaml
source_id: payroll_csv

entities:
  crm_user:
    from: payroll.csv
    id: user_id
    fields:
      user_id: user
      full_name: full_name

from is the filename (not the full path — the path lives in the profile only). Map playbook fields to CSV column headers when names differ. Introspect reads column headers from the file: introspect_source with source_id: payroll_csv.

MongoDB

1. Set credentials in `.env`

AG_MONGODB_DSN=mongodb://localhost:27017
AG_MONGODB_DATABASE=your_database_name

2. Register in `profiles/local.yaml`

sources:
  mongo_main:
    adapter: mongodb
    dsn: env:AG_MONGODB_DSN
    database: env:AG_MONGODB_DATABASE

3. Use in a binding

# bindings/your-playbook.mongodb.yaml
source_id: mongo_main

entities:
  crm_customer:
    from: customers
    id: customer_id
    fields:
      customer_id: _id
      full_name: name

from is the collection name. Field values are document field paths. Introspect with introspect_source, source_id: mongo_main, and schema_name set to the database name if not set in the profile.

For a full Postgres + MongoDB walkthrough, see Cross-source CRM walkthrough.

Verify your connection

After editing .env and profiles/local.yaml, restart the stack, then:

CLI health

anythinggraph doctor
anythinggraph status

MCP (admin token)

From Cursor or another MCP client connected to http://127.0.0.1:3334/mcp:

list_sources — shows configured profile keys and adapter types (no secrets)
introspect_source — schema, tables/collections, or Salesforce objects
sample_source — a few live rows or documents to confirm connectivity

If a source does not appear in list_sources, check YAML syntax, restart the service, and confirm AG_PROFILE_PATH points at the file you edited.

What comes next

Bindings — map playbook entities to tables, collections, or objects
Playbooks & bindings guide — step-by-step authoring
MCP setup — connect Claude or Cursor
All data adapters — Postgres, REST, SQL Server, and planned connectors

Connect your data sources

Where is local.yaml?

Profile structure

How profiles work

Quick checklist (every source)

MySQL

1. Set credentials in .env

2. Register in profiles/local.yaml

3. Use in a binding

Salesforce

1. Set credentials in .env

2. Register in profiles/local.yaml

3. Use in a binding

CSV files

1. Set file path in .env

2. Register in profiles/local.yaml

3. Use in a binding

MongoDB

1. Set credentials in .env

2. Register in profiles/local.yaml

3. Use in a binding

Verify your connection

CLI health

MCP (admin token)

What comes next

Where is `local.yaml`?

1. Set credentials in `.env`

2. Register in `profiles/local.yaml`

1. Set credentials in `.env`

2. Register in `profiles/local.yaml`

1. Set file path in `.env`

2. Register in `profiles/local.yaml`

1. Set credentials in `.env`

2. Register in `profiles/local.yaml`