in-toto is a framework to protect supply chain integrity.
in-toto provides a framework to protect the integrity of the software supply chain. It does so by verifying that each task in the chain is carried out as planned, by authorized personnel only, and that the product is not tampered with in transit.
in-toto requires a project owner to create a layout. A layout lists the sequence of steps of the software supply chain, and the functionaries authorized to perform these steps. When a functionary performs a step in-toto gathers information about the used command and the related files and stores it in a link metadata file. As a consequence link files provide the required evidence to establish a continuous chain that can be validated against the steps defined in the layout.
The layout, signed by the project owners, together with the links, signed by the designated functionaries, are released as part of the final product, and can be validated manually or via automated tooling in, e.g. a package manager.
in-toto is available on PyPI and can be installed viain-toto.readthedocs.io to learn about system dependencies and installation alternatives and recommendations.
pip install in-toto
The in-toto software supply chain layout consists of the following parts: - expiration date - readme (an optional description of the supply chain) - functionary keys (public keys, used to verify link metadata signatures) - signatures (one or more layout signatures created with the project owner key(s)) - software supply chain steps correspond to steps carried out by a functionary as part of the software supply chain. The steps defined in the layout list the functionaries who are authorized to carry out the step (by key id). Steps require a unique name to associate them (upon verification) with link metadata that is created when a functionary carries out the step using the
in-tototools. Additionally, steps must have material and product rules which define the files a step is supposed to operate on. Material and product rules are described in the section below. - inspections define commands to be run during the verification process and can also list material and product rules.
A software supply chain usually operates on a set of files, such as source code, executables, packages, or the like. in-toto calls these files artifacts. A material is an artifact that will be used when a step or inspection is carried out. Likewise, a product is an artifact that results from carrying out a step.
The in-toto layout provides a simple rule language to authorize or enforce the artifacts of a step and to chain them together. This adds the following guarantees for any given step or inspection: - Only artifacts authorized by the project owner are created, modified or deleted, - each defined creation, modification or deletion is enforced, and also - restricted to the scope of its definition, which chains subsequent steps and inspections together.
Note that it is up to you to properly secure your supply chain, by authorizing, enforcing and chaining materials and products using any and usually multiple of the following rules: -
MATCH [IN ] WITH (MATERIALS|PRODUCTS) [IN ] FROM
Rule arguments specified as
allow for Unix shell-style wildcards as implemented by Python's
To learn more about the different rule types, their guarantees and how they are applied take a look at the Artifact Rules section of the in-toto specification.
in-toto-runis used to execute a step in the software supply chain. This can be anything relevant to the project such as tagging a release with
git, running a test, or building a binary. The relevant step name and command are passed as arguments, along with materials, which are files required for that step's command to execute, and products which are files expected as a result of the execution of that command. These, and other relevant details pertaining to the step are stored in a link file, which is signed using the functionary's key.
If materials are not passed to the command, the link file generated just doesn't record them. Similarly, if the execution of a command via
in-toto-rundoesn't result in any products, they're not recorded in the link file. Any files that are modified or used in any way during the execution of the command are not recorded in the link file unless explicitly passed as artifacts. Conversely, any materials or products passed to the command are recorded in the link file even if they're not part of the execution of the command.
See this simple usage example from the demo application for more details. For a detailed list of all the command line arguments, run
in-toto-run --helpor look at the online documentation.
in-toto-recordworks similar to
in-toto-runbut can be used for multi-part software supply chain steps, i.e. steps that are not carried out by a single command. Use
in-toto-record start ...to create a preliminary link file that only records the materials, then run the commands of that step or edit files manually and finally use
in-toto-record stop ...to record the products and generate the actual link metadata file. For a detailed list of all command line arguments and their usage, run
in-toto-record start --helpor
in-toto-record stop --help, or look at the online documentation.
In order to verify the final product with in-toto, the verifier must have access to the layout, the
*.linkfiles, and the project owner's public key(s).
in-toto-verifyon the final product to verify that - the layout was signed with the project owner's private key(s), - has not expired, - each step was performed and signed by the authorized functionary, - the functionaries used the commands, they were supposed to use, - materials and products of each step were in place as defined by the rules, and - run the defined inspections
For a detailed list of all command line arguments and their usage, run
in-toto-verify --helpor look at the online documentation.
in-toto-signis a metadata signature helper tool to add, replace, and verify signatures within in-toto Link or Layout metadata, with options to: - replace (default) or add signature(s), with layout metadata able to be signed by multiple keys at once while link metadata can only be signed by one key at a time - write signed metadata to a specified path (if no output path is specified, layout metadata is written to the path of the input file while link metadata is written to
..link) - verify signatures
This tool is intended to sign layouts created by the layout web wizard, but also serves well to re-sign test and demo data. For example, it can be used if metadata formats or signing routines change.
For a detailed list of all command line arguments and their usage, run
in-toto-sign --helpor look at the online documentation.
You can try in-toto by running the demo application. The demo basically outlines three users viz., Alice (project owner), Bob (functionary) and Carl (functionary) and how in-toto helps to specify a project layout and verify that the layout has been followed in a correct manner.
You can read more about how in-toto works by taking a look at the specification.
Security issues can be reported by emailing [email protected]
At a minimum, the report must contain the following: * Description of the vulnerability. * Steps to reproduce the issue.
Optionally, reports that are emailed can be encrypted with PGP. You should use PGP key fingerprint E9C0 59EC 0D32 64FA B35F 94AD 465B F9F6 F8EB 475A.
Please do not use the GitHub issue tracker to submit vulnerability reports. The issue tracker is intended for bug reports and to make feature requests.
Development of in-toto occurs on the "develop" branch of this repository. Contributions can be made by submitting GitHub Pull Requests. Take a look at our development guidelines for detailed instructions. Submitted code should follow our style guidelines and must be unit tested.
Contributors must also indicate acceptance of the Developer Certificate of Origin (DCO) by appending a
Signed-off-by: Your Nameto each git commit message (see
git commit --signoff).
This project is managed by Prof. Santiago Torres-Arias at Purdue University. It is worked on by many folks in academia and industry, including members of the Secure Systems Lab at NYU and the NJIT Cybersecurity Research Center.
This research was supported by the Defense Advanced Research Projects Agency (DARPA), the Air Force Research Laboratory (AFRL), and the US National Science Foundation (NSF). Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of DARPA, AFRL, and NSF. The United States Government is authorized to reproduce and distribute reprints notwithstanding any copyright notice herein.