Current version: 0.2.1 Alpha

Arboric is the first, and so far only API proxy / gateway dedicated specifically for GraphQL. It aims to provide several key features:

Currently supports logging of query and mutation fields and counts to InfluxDB. In the near future, Arboric aims to:

• allow selectively logging requests metadata such as JWT claims & values
• support logging to Kafka

Currently, Arboric can enforce verification of a JWT Authorization: Bearer token using a supplied HS256 signing key (via environment variable). In the near future, it aims to support:

• Supplying the HS256 signing key as a hex, base64, or 'raw' value from the environment, directly as run-time argument or configuration value, or from a file
• Support for RS256 asymmetric token verification

Arboric provides Attribute Based Access Control that allows great flexibility in access controls. Currently, it supports matching:

• JWT claim presence
• JWT claim equality
• JWT claim inclusion (e.g. claims["roles"] includes "admin" will match "roles": "user, admin")

It also supports Allow or Deny rules based on GraphQL pattern matching. For example:

• foo or query:foo matches a query for the field foo
• mutation:doSomething matches the mutation doSomething
• * or query:* matches any query, while
• mutation:* matches any mutation

In the future, Arboric aims to allow:

• nested fields matching, e.g.
  • hero.secretIdentity
  • hero.friends.secretIdentity
  • **.secretIdentity
• matching by GraphQL type, e.g. type:Hero
• matching by type and field, e.g. type:Hero{secretIdentity}

In /etc/arboric/config.yml

arboric:
  log:
    console:
      level: info
listeners:
- bind: localhost
  port: 4000
  proxy: http://localhost:3001/graphql
  jwt_signing_key:
    from_env:
      key: SECRET_KEY_BASE
      encoding: hex
  log_to:
    influx_db:
      uri: http://localhost:8086
      database: arboric
  policies:
  - when:
    - claim_is_present: sub
    allow:
    - query: "*"
    deny:
    - query: "__*"
    - mutation: "*"
  - when:
    - claim: roles
      includes: admin
    allow:
    - "*"

The above configuration translates as follows:

• listen on localhost at port 4000
• forward requests to http://localhost:3000/graphql
• require a valid JWT Authorization: Bearer token, signed using the hexadecimal key in the environment variable SECRET_KEY_BASE
• log requests (queries & mutations) to InfluxDB at http://localhost:8086 to the arboric databaes

Additionally, attribute-based access control policies specify that:

• an authenticated caller (with a sub claim) can execute any query except those beginning with __ (the GraphQL introspection queries), and cannot execute any mutations, but
• a caller whose roles claim (a comma-separated list) includes admin can execute any query or mutation

• TLS/SSL edge termination
• Two-way TLS certificate authentication/validation from edge to backend
• Multiple listeners on a single server process

Currently, Arboric is not yet distributed in binary. In the future, we intend to make Arboric available as binary packages for Linux and Mac OS, as well as a Docker image.

To build Arboric requires:

• Rust 1.37.0

git clone https://gitlab.com/arboric/arboric

cargo build --release

./target/release/arboric -f /etc/arboric/config.yml start

Given the above configuration file, this will start Arboric listening on port 4000, forwarding requests to http://localhost:3001/graphql, and validating the JWT Authorization: Bearer token using the SECRET_KEY_BASE (hexadecimal) environment variable value, and so forth.

• Allow for multiple Listeners
• Arboric API (in GraphQL, of course)
• Allow for run-time configuration (via the API)

• Configurable log level
• File logger
• Configurable policies
• Configurable InfluxDB logging

• ABAC (Attribute Based Access Control), allows for
  • Matching by claim presence, equality, or inclusion
• Configuration model
• Read configuration from YAML
• CLI arguments processor

• JWT Authentication
• Logging to InfluxDB
• Role and Path-based Access Control Lists (black/white list)