Management and Distribution of Secret Credentials
Switch branches/tags
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.


Build Status Hackage Version


Warning: This is an experimental prototype/preview release which is still in development and not intended for public consumption, caveat emptor!

The credentials library and CLI provides a unified interface for managing secure, shared credentials.

It uses Key Management Service (KMS) for master key management, locally encrypts and decrypts secrets, which are then stored in any of the supported storage backends. (Currently DynamoDB.)

The use-case is to avoid storing sensitive information such as passwords and connection strings in plaintext in places such as source control or on developer machines. Instead you can securely administer and distribute secrets, leveraging Amazon's IAM policies for access control and permissions to ensure limited read-only permissions from production/deployed hosts where applicable.

Please see the introductory blog post for more information, or the Haddock documentation built by CI from the develop branch, which can be found here.

If Haskell is not your thing, check out credstash, the Python project that inspired credentials.


  1. stack install credentials-cli
  2. Create a new key in KMS called "credentials". You can do this under Identity and Access Management > Encryption Keys in the AWS developer console.
  3. Your AWS access credentials are available where amazonka can find them. This will be automatic if you are running on an EC2 host, otherwise the ~/.aws/credentials file or AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY environment variables need to be configured.
  4. credentials setup


The following is an example of using the credentials library as a dependency of your Haskell project. It retrieves a database connection string containing a sensitive password, when a webserver starts. It's worth pointing out the setup all pertains to the underlying amazonka library, since all of the credentials operations run in a MonadAWS context.

{-# LANGUAGE OverloadedStrings #-}

import Control.Lens

import Credentials

import Data.ByteString (ByteString)

import Network.AWS
import Network.Wai              (Application)
import Network.Wai.Handler.Warp (run)

import System.IO (stdout)

main :: IO
main = do
    -- A new 'Logger' to replace the default noop logger is created,
    -- which will print AWS debug information and errors to stdout.
    lgr <- newLogger Debug stdout

    -- A new amazonka 'Env' is created, which auto-discovers the
    -- underlying host credentials.
    env <- newEnv Frankfurt Discover

    let table = Credentials.defaultTable
        key   = Credentials.defaultKeyId
        name  = "secret-database-uri"

    -- We now run the 'AWS' computation with the overriden logger,
    -- performing sequence credentials operation(s).
    -- For 'select', the plaintext and corresponding revision is returned.
    (uri, _) <- runResourceT . runAWS (env & envLogger .~ lgr) $ do
        -- Selecting the credential by name, and specifying 'Nothing' for the
        -- revision results in the latest revision of the credential. mempty name Nothing table

    -- We can now connect to the database using our sensitive connection URI.
    run 3000 (app uri)

app :: ByteString -> Application
app uri rq f = ...

A basic example of using the CLI is as follows:

$ credentials setup
Setting up dynamo:///credentials in eu-central-1.
Running ...
  status: created
$ credentials insert --name foo --secret "A magical secret."
Writing new revision of foo to dynamo:///credentials in eu-central-1...
  name: foo
  revision: 82687c4
$ credentials select --name foo
Retrieving foo from dynamo:///credentials in eu-central-1...
  name: foo
  revision: 82687c4
  secret: A magical secret.
$ credentials list
Listing contents of dynamo:///credentials in eu-central-1...
    - 82687c4 # latest

Additional means of formatting the output is available, for example the --format option supports json (JSON formatting) or echo (single shell value) output.

Credential Contexts

KMS Encryption Context is optionally supported, which is used to check ciphertext integrity but not stored as part of the encryption parameters.

This means you need to pass the exact context that was used for encryption, when decrypting, and allows the use of KMS Key Policy and KMS Grant conditions. One such use would be to ensure your web and database servers can read a database password, but your database servers cannot read your web server's private key.

For example, supplying encryption context would look as follows:

$ credentials insert --name hello --secret world \
    --context application=db \
    --context environment=production

$ credentials select --name hello \
    --context application=db \
    --context environment=production

Failure to specify the exact encryption context resullts in an informative error during decryption.

Credential Revisions

Credential storage is immutable and the insertion of a credential into storage is append-only. When a value for an existing credential name is inserted a unique revision is generated which you can use in combination with the name to uniquely address any credential in the system.

Rotation of credentials can be performed by simply inserting a new value for a given credential name, and any system which selects that credential without a specific revision will always receive the latest revision's value.

You can also specify the exact revision when selecting a credential to perform pinning.

IAM Policies

The complete list of AWS KMS and DynamoDB permissions the various operations will require are:

  • Setup and teardown:
    • CreateTable
    • DestroyTable
    • ListTables
  • List credentials and revisions:
    • Scan
  • Insert a new encrypted credential revision:
    • GenerateDataKey
    • Query
    • PutItem
  • Select and decrypt an existing credential revision:
    • Decrypt
    • Query
  • Deletion of a credential revision, or revisions:
    • Query
    • DeleteItem

It's recommended you allow only the minimal set of permissions as your usecase requires. For example, the following two IAM policies illustrate the minimum allowable API operations for read/write, and read only access respectively:

Read and Write

  "Version": "2012-10-17",
  "Statement": [
      "Action": [
      "Effect": "Allow",
      "Resource": "arn:aws:kms:us-east-1:AWS_ACCOUNT_ID:alias/credentials"
      "Action": [
      "Effect": "Allow",
      "Resource": "arn:aws:dynamodb:us-east-1:AWS_ACCOUNT_ID:table/credentials"

Read Only

  "Version": "2012-10-17",
  "Statement": [
      "Action": [
      "Effect": "Allow",
      "Resource": "arn:aws:kms:us-east-1:AWS_ACCOUNT_ID:alias/credentials"
      "Action": [
      "Effect": "Allow",
      "Resource": "arn:aws:dynamodb:us-east-1:AWS_ACCOUNT_ID:table/credentials"

Remember to replace the use of AWS_ACCOUNT_ID above with your own account identifier.

Security Considerations

Any IAM user who can read items from the DynamoDB credentials table and can call KMS Decrypt, can obtain the plaintext of stored credentials.

Similarly when using credentials as part of an EC2 deployment, the instance boundary will be the security boundary as the IAM role credentials are available via the local EC2 metadata service at If your instance is compromised and has an IAM role assigned that similarly allows reading items from the DynamoDB table and calling KMS Decrypt, the attacker will likewise be able to recover credentials.

If the instance boundary is too coarse, consider using iptables or similar to restrict metadata access to privileged users.

Service Pricing

A single master key in KMS costs $1 USD per month. The DynamoDB table throughput is configured to use 1 provisioned read and 1 provisioned write, so if you are using less than the free tier limit of 25 reads and 25 writes per second, only the KMS charges will apply.

If you are likely to utilise much more than 25 reads/writes per second, you can estimate your monthly charges by using the AWS pricing calculator.

TL;DR, $1 USD per month for the predicted usecase.

Local Development

You can use both the library and CLI in a limited offline fashion by running a DynamoDBLocal instance using the script/ script. The CLI commands can then take a --uri dynamo://localhost:4000/credentials parameter to force communication with the local DynamoDB database.

Unfortunately there exists no similar mechanism to run KMS locally, currently.


For any problems, comments, or feedback please create an issue here on GitHub.


credentials is released under the Apache License Version 2.0.