A framework to define and secure the integrity of software supply chains
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 via pip
. See
in-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:
in-toto
tools. 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.Take a look at the demo layout creation example for further information on how to create an in-toto layout. Or try our experimental layout creation web tool.
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:
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:
CREATE <pattern>
DELETE <pattern>
MODIFY <pattern>
ALLOW <pattern>
DISALLOW <pattern>
REQUIRE <file>
MATCH <pattern> [IN <source-path-prefix>] WITH (MATERIALS|PRODUCTS) [IN <destination-path-prefix>] FROM <step>
Rule arguments specified as <pattern>
allow for Unix shell-style wildcards as implemented by Python's fnmatch
.
in-toto's Artifact Rules, by default, allow artifacts to exist if they are not explicitly disallowed. As such, a DISALLOW *
invocation is recommended as the final rule for most step definitions. 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-run
is 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-run
doesn'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 --help
or look at the online
documentation.
in-toto-record
works similar to in-toto-run
but 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 --help
or 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 *.link
files,
and the project owner's public key(s).
Use in-toto-verify
on the final product to verify that
For a detailed list of all command line arguments and their usage, run
in-toto-verify --help
or look at the
online
documentation.
in-toto-sign
is a metadata signature helper tool to add, replace, and
verify signatures within in-toto Link or Layout metadata, with options to:
<name>.<keyid prefix>.link
)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 --help
or 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.
See SECURITY.md.
For information about in-toto's governance and contributing guidelines, see GOVERNANCE.md.
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.