Chainloop: Open-source supply chain evidence & policy engine

A software supply chain produces a mountain of evidence, but none of it means anything if it cannot be tied back to a specific build. Chainloop, an open-source evidence store, solves this by capturing what a pipeline creates and sealing it inside a signed, verifiable record. The tool runs as a command line utility inside popular CI systems like GitHub Actions, GitLab, Jenkins, or Dagger, collects the build outputs, uploads them to content-addressable storage, and references each artifact in a signed in-toto attestation. This specification tracks exactly who ran which step, making the entire process auditable after the fact.

For compliance and security teams, all that data lands in a centralized control plane, already signed, regardless of which CI provider produced it. The problem this solves becomes obvious when you look at a typical pipeline without it. A build generates a bill of materials listing every dependency, a SARIF file from static analysis, a coverage report, a container image reference, and a scan from a commercial security tool. Each of those files ends up in a different location, with nothing linking them to the originating commit. Six months later, when a customer asks which version of a compression library shipped in a particular release, someone is stuck digging through build logs to reconstruct an answer they cannot verify.

The enforcement comes from a Workflow Contract, which declares what a build must deliver: the required materials, build information, and the environment it has to run in. Compliance teams write the contract, and Chainloop checks that the artifact creation and attestation meet those requirements. If a pipeline stops producing a bill of materials, it stops satisfying the contract. Policies written in Rego, the language used by Open Policy Agent, attach to those contracts. They are evaluated automatically, and the results are embedded into the attestation before signing and storage, so the verdict travels inside the signed record instead of living in a dashboard that can be altered later.

The tool handles seventeen named evidence formats out of the box, with a longer catalog available. That includes CycloneDX and SPDX bills of materials, OpenVEX, all four CSAF document types, SARIF, ZAP DAST results, BlackDuck SCA output, PrismaCloud Twistcli scans, GitLab security reports, JUnit results, JaCoCo XML coverage, Helm charts, and container image references. Anything else fits into one of four catch-all categories, whether as a custom evidence type, a JSON approval report, or key-value metadata pairs.

Signing is flexible too. Evidence can be authenticated through Sigstore or an organization’s own PKI, including AWS KMS or Keyfactor, which matters for teams that must keep keys in-house.

Artifacts and evidence can be routed to an OCI registry or cloud blob storage, sent to Dependency-Track or Guac for bill-of-materials analysis, and announced in Jira, Discord, or Slack. Swapping the analysis backend requires no pipeline changes, because the pipeline only ever talks to the crafting tool. Compliance and security teams define contracts, write policies, set up integrations, and control access to the control plane. Development teams get a simpler pitch: just run the tool and let it handle the rest.

Regulatory pressure is driving much of the demand. FedRAMP, Executive Order 14028, the EU Cyber Resilience Act, and the Digital Operational Resilience Act all turn questions about what shipped into questions someone has to answer on a deadline. The project publishes guides for the Cyber Resilience Act and for SLSA, the supply chain framework whose level 3 the single source of truth is designed to meet. A FedRAMP guide is listed as forthcoming.

The command line tool defaults to a hosted Chainloop instance, which is convenient for a trial but means evidence leaves your infrastructure. Running the Helm chart on your own Kubernetes cluster brings the control plane back in-house. The project is available for free on GitHub.