Talisman

A tool to detect and prevent secrets from getting checked in

Table of Contents

• What is Talisman?
• Installation
  • As a global hook template (Recommended)
  • To a single repository
• Upgrading Talisman
• Talisman in action
  • Validations
  • Ignoring files
  • Talisman as a CLI utility
    • Git History Scanner
    • Checksum Calculator
  • Talisman HTML Reporting
• Uninstallation
  • From a global hook template
  • From a single repository
• Contributing to Talisman
  • Developing locally
  • Releasing

What is Talisman?

Talisman is a tool that installs a hook to your repository to ensure that potential secrets or sensitive information do not leave the developer's workstation.

It validates the outgoing changeset for things that look suspicious - such as potential SSH keys, authorization tokens, private keys etc.

Installation

Talisman supports MAC OSX, Linux and Windows.

Talisman can be installed and used in one of the following ways:

1. As a git hook as a global git hook template
2. As a git hook into a single git repository

Talisman can be set up as either a pre-commit or pre-push hook on the git repositories.

Find the instructions below.

[Recommended approach]

Installation as a global hook template

We recommend installing Talisman as a pre-commit git hook template, as that will cause Talisman to be present, not only in your existing git repositories, but also in any new repository that you 'init' or 'clone'.

1. Run the following command on your terminal, to download and install the binary at $HOME/.talisman/bin

As a pre-commit hook:

curl --silent  https://raw.githubusercontent.com/thoughtworks/talisman/master/global_install_scripts/install.bash > /tmp/install_talisman.bash && /bin/bash /tmp/install_talisman.bash

OR

As a pre-push hook:

curl --silent  https://raw.githubusercontent.com/thoughtworks/talisman/master/global_install_scripts/install.bash > /tmp/install_talisman.bash && /bin/bash /tmp/install_talisman.bash pre-push

2. If you do not have TALISMAN_HOME set up in your $PATH, you will be asked an appropriate place to set it up. Choose the option number where you set the profile source on your machine.

Remember to execute source on the path file or restart your terminal. If you choose to set the $PATH later, please export TALISMAN_HOME=$HOME/.talisman/bin to the path.

3. Choose a base directory where Talisman should scan for all git repositories, and setup a git hook (pre-commit or pre-push, as chosen in step 1) as a symlink. This script will not clobber pre-existing hooks. If you have existing hooks, [look for ways to chain Talisman into them.] (#handling-existing-hooks)

Handling existing hooks

Installation of Talisman globally does not clobber pre-existing hooks on repositories.
If the installation script finds any existing hooks, it will only indicate so on the console.
To achieve running multiple hooks we suggest (but not limited to) the following two tools

1. Pre-commit (Linux/Unix)

Use pre-commit tool to manage all the existing hooks along with Talisman. In the suggestion, it will prompt the following code to be included in .pre-commit-config.yaml

    -   repo: local
        hooks:
        -   id: talisman-precommit
            name: talisman
            entry: bash -c 'if [ -n "${TALISMAN_HOME:-}" ]; then ${TALISMAN_HOME}/talisman_hook_script pre-commit; else echo "TALISMAN does not exist. Consider installing from https://github.com/thoughtworks/talisman . If you already have talisman installed, please ensure TALISMAN_HOME variable is set to where talisman_hook_script resides, for example, TALISMAN_HOME=${HOME}/.talisman/bin"; fi'
            language: system
            pass_filenames: false
            types: [text]
            verbose: true

2. Husky (Linux/Unix/Windows)

husky is an npm module for managing git hooks. In order to use husky, make sure you have already set TALISMAN_HOME to $PATH.

• Existing Users

If you already are using husky, add the following lines to husky pre-commit in package.json

Windows

   "bash -c '\"%TALISMAN_HOME%\\${TALISMAN_BINARY_NAME}\" --githook pre-commit'"

Linux/Unix

   $TALISMAN_HOME/talisman_hook_script pre-commit

• New Users

If you want to use husky with multiple hooks along with talisman, add the following snippet to you package json.

Windows

    {
       "husky": {
         "hooks": {
           "pre-commit": "bash -c '\"%TALISMAN_HOME%\\${TALISMAN_BINARY_NAME}\" --githook pre-commit'" && "other-scripts"
           }
       }
   }

Linux/Unix

   {
     "husky": {
      "hooks": {
        "pre-commit": "$TALISMAN_HOME/talisman_hook_script pre-commit" && "other-scripts"
         }
       }
     }

Installation to a single project

# Download the talisman binary
curl https://thoughtworks.github.io/talisman/install.sh > ~/install-talisman.sh
chmod +x ~/install-talisman.sh

# Install to a single project (as pre-push hook)
cd my-git-project
~/install-talisman.sh

Handling existing hooks

Talisman will need to be chained with any existing git hooks.You can use pre-commit git hooks framework to handle this.

Add this to your .pre-commit-config.yaml (be sure to update rev to point to a real git revision!)

-   repo: https://github.com/thoughtworks/talisman
    rev: ''  # Update me!
    hooks:
    # either `commit` or `push` support
    -   id: talisman-commit
    # -   id: talisman-push

Upgrading

Since release v0.4.4, Talisman automatically updates the binary to the latest release, when the hook is invoked (at pre-commit/pre-push, as set up). So, just sit back, relax, and keep using the latest Talisman without any extra efforts.

If at all you need to manually upgrade, here are the steps:
[Recommended] Update Talisman binary and hook scripts to the latest release:

curl --silent  https://raw.githubusercontent.com/thoughtworks/talisman/master/global_install_scripts/update_talisman.bash > /tmp/update_talisman.bash && /bin/bash /tmp/update_talisman.bash

Update only Talisman binary by executing:

curl --silent  https://raw.githubusercontent.com/thoughtworks/talisman/master/global_install_scripts/update_talisman.bash > /tmp/update_talisman.bash && /bin/bash /tmp/update_talisman.bash talisman-binary

Talisman in action

After the installation is successful, Talisman will run checks for obvious secrets automatically before each commit or push (as chosen during installation). In case there are any security breaches detected, talisman will display a detailed report of the errors:

$ git push
Talisman Report:
+-----------------+-------------------------------------------------------------------------------+
|     FILE        |                                    ERRORS                                     |
+-----------------+-------------------------------------------------------------------------------+
| danger.pem      | The file name "danger.pem"                                                    |
|                 | failed checks against the                                                     |
|                 | pattern ^.+\.pem$                                                             |
+-----------------+-------------------------------------------------------------------------------+
| danger.pem      | Expected file to not to contain hex encoded texts such as:                    |
|                 | awsSecretKey=c64e8c79aacf5ddb02f1274db2d973f363f4f553ab1692d8d203b4cc09692f79 |
+-----------------+-------------------------------------------------------------------------------+

In the above example, the file danger.pem has been flagged as a security breach due to the following reasons:

• The filename matches one of the pre-configured patterns.
• The file contains an awsSecretKey which is scanned and flagged by Talisman

If you have installed Talisman as a pre-commit hook, it will scan only the diff within each commit. This means that it would only report errors for parts of the file that were changed.

In case you have installed Talisman as a pre-push hook, it will scan the complete file in which changes are made. As mentioned above, it is recommended that you use Talisman as a pre-commit hook.

Validations

The following detectors execute against the changesets to detect secrets/sensitive information:

• Encoded values - scans for encoded secrets in Base64, hex etc.
• File content - scans for suspicious content in file that could be potential secrets or passwords
• File size - scans for large files that may potentially contain keys or other secrets
• Entropy - scans for content with high entropy that are likely to contain passwords
• Credit card numbers - scans for content that could be potential credit card numbers
• File names - scans for file names and extensions that could indicate them potentially containing secrets, such as keys, credentials etc.

Ignoring Files

If you're really sure you want to push that file, you can configure it into the .talismanrc file in the project root. The contents required for ignoring your failed files will be printed by Talisman on the console immediately after the Talisman Error Report:

If you are absolutely sure that you want to ignore the above files from talisman detectors, consider pasting the following format in .talismanrc file in the project root
fileignoreconfig:
- filename: danger.pem
  checksum: cf97abd34cebe895417eb4d97fbd7374aa138dcb65b1fe7f6b6cc1238aaf4d48
  ignore_detectors: []

Entering this in the .talismanrc file will ensure that Talisman will ignore the danger.pem file as long as the checksum matches the value mentioned in the checksum field.

Ignoring specific detectors

Below is a detailed description of the various fields that can be configured into the .talismanrc file:

• filename : This field should mention the fully qualified filename.
• checksum : This field should always have the value specified by Talisman in the message displayed above. If at any point, a new change is made to the file, it will result in a new checksum and Talisman will scan the file again for any potential security threats.
• ignore_detectors : This field will disable specific detectors for a particular file. For example, if your init-env.sh filename triggers a warning, you can only disable this warning while still being alerted if other things go wrong (e.g. file content):

fileignoreconfig:
- filename: init-env.sh
  checksum: cf97abd34cebe895417eb4d97fbd7374aa138dcb65b1fe7f6b6cc1238aaf4d48
  ignore_detectors: [filename, filesize]

Note: Here both filename and filesize detectors are ignored for init-env.sh, but filecontent detector will still activate on init-env.sh

At the moment, you can ignore

• filecontent
• filename
• filesize

Ignoring multiple files of same type (with wildcards)

You can choose to ignore all files of a certain type, because you know they will always be safe, and you wouldn't want Talisman to scan them.

Steps:

1. Format a wildard pattern for the files you want to ignore. For example, *.lock
2. Use the checksum calculator to feed the pattern and attain a collective checksum. For example, talisman --checksum="*.lock"
3. Copy the fileconfig block, printed on console, to .talismanrc file.

If any of the files are modified, talisman will scan the files again, unless you re-calculate the new checksum and replace it in .talismanrc file.

Note: The use of .talismanignore has been deprecated. File .talismanrc replaces it because:

• .talismanrc has a much more legible yaml format
• It also brings in more secure practices with every modification of a file with a potential sensitive value to be reviewed
• The new format also brings in the extensibility to introduce new usable functionalities. Keep a watch out for more

Talisman as a CLI utility

If you execute talisman on the command line, you will be able to view all the parameter options you can pass

	  --c string          short form of checksum calculator
     --checksum string    checksum calculator calculates checksum and suggests .talsimarc format
      --d                 short form of debug
      --debug             enable debug mode (warning: very verbose)
      --githook string    either pre-push or pre-commit (default "pre-push")
      --p string          short form of pattern
      --pattern string    pattern (glob-like) of files to scan (ignores githooks)
      --s                 short form of scanner
      --scan              scanner scans the git commit history for potential secrets
      --v                 short form of version
      --version           show current version of talisman

Git history Scanner

You can now execute Talisman from CLI, and potentially add it to your CI/CD pipelines, to scan git history of your repository to find any sensitive content. This includes scanning of the files listed in the .talismanrc file as well.

Steps:

1. Get into the git directory path to be scanned cd <directory to scan>
2. Run the scan command talisman --scan

• Running this command will create a folder named talisman_reports in the root of the current directory and store the report files there.
• You can also specify the location for reports by providing an additional parameter as --reportDirectory or --rd
  For example, talisman --scan --reportdirectory=/Users/username/Desktop

You can use the other options to scan as given above.

Talisman currently does not support ignoring of files for scanning.

Checksum Calculator

Talisman Checksum calculator gives out yaml format which you can directly copy and paste in .talismanrc file in order to ignore particular file formats from talisman detectors.

To run the checksum please "cd" into the root of your repository and run the following command

For Example: talisman --checksum="*.pem *.txt"

1. This command finds all the .pem files in the respository and calculates collective checksum of all those files and outputs a yaml format for .talismanrc. In the same way it deals with the .txt files.
2. Multiple file names / patterns can be given with space seperation.

Example output:

.talismanrc format for given file names / patterns
fileignoreconfig:
- filename: '*.pem'
  checksum: f731b26be086fd2647c40801630e2219ef207cb1aacc02f9bf0559a75c0855a4
  ignore_detectors: []
- filename: '*.txt'
  checksum: d9e9e94868d7de5b2a0706b8d38d0f79730839e0eb4de4e9a2a5a014c7c43f35
  ignore_detectors: []

Note: Checksum calculator considers the staged files while calculating the collective checksum of the files.

Talisman HTML Reporting

Powered by

Talisman CLI tool talisman also comes with the capability to provide detailed and sharable HTML report. Once you have installed Talisman, please follow the steps mentioned in talisman-html-report, to install the reporting package in .talisman folder. To generate the html report, run:

• talisman --scanWithHtml

This will scan the repository and create a folder talisman_html_report under the the scanned repository. We need to start an HTTP server inside this repository to access the report.Below is a recommended approach to start a HTTP server:

• python -m SimpleHTTPServer <port> (eg: 8000)

You can now access the report by navigating to:

http://localhost:8000

Sample Screenshots

• Welcome

• Summary

• Detailed Report

• Error Report

Note: You don't have to start a server if you are running Talisman in CI or any other hosted environment

Uninstallation

The uninstallation process depends on how you had installed Talisman. You could have chosen to install as a global hook template or at a single repository.

Please follow the steps below based on which option you had chosen at installation.

Uninstallation from a global hook template

Run the following command on your terminal to uninstall talisman globally from your machine.

For pre-commit hook:

curl --silent  https://raw.githubusercontent.com/thoughtworks/talisman/master/global_install_scripts/uninstall.bash > /tmp/uninstall_talisman.bash && /bin/bash /tmp/uninstall_talisman.bash

For pre-push hook:

curl --silent  https://raw.githubusercontent.com/thoughtworks/talisman/master/global_install_scripts/uninstall.bash > /tmp/uninstall_talisman.bash && /bin/bash /tmp/uninstall_talisman.bash pre-push

This will

1. ask you for the base dir of all your repos, find all git repos inside it and remove talisman hooks
2. remove talisman hook from .git-template
3. remove talisman from the central install location ($HOME/.talisman/bin).
   

You will have to manually remove TALISMAN_HOME from your environment variables

Uninstallation from a single repository

When you installed Talisman, it must have created a pre-commit or pre-push hook (as selected) in your repository during installation.

You can remove the hook manually by deleting the Talisman pre-commit or pre-push hook from .git/hooks folder in repository.

Contributing to Talisman

Developing locally

To contribute to Talisman, you need a working golang development environment. Check this link to help you get started with that.

Talisman now uses go modules (GO111MODULE=on) to manage dependencies

Once you have go 1.11 installed and setup, clone the talisman repository. In your working copy, fetch the dependencies by having go mod fetch them for you.

GO111MODULE=on go mod vendor

To run tests GO111MODULE=on go test -mod=vendor ./...

To build Talisman, we can use gox:

gox -osarch="darwin/amd64 linux/386 linux/amd64"

Convenince scripts ./build and ./clean perform build and clean-up as mentioned above.

Releasing

• Follow the instructions at the end of 'Developing locally' to build the binaries
• Bump the version in install.sh according to semver conventions
• Update the expected hashes in install.sh to match the new binaries you just created (shasum -b -a256 ...)
• Make release commit and tag with the new version prefixed by v (like git tag v0.3.0)
• Push your release commit and tag: git push && git push --tags
• Create a new release in github, filling in the new commit tag you just created
• Update the install script hosted on github pages: git checkout gh-pages, git checkout master -- install.sh, git commit -m ...

The latest version will now be accessible to anyone who builds their own binaries, downloads binaries directly from github releases, or uses the install script from the website.