Software supply-chain security
Know when the software you trust changes.
SCG records dependency fingerprints and flags changes against your baseline. Run checks on your own computer, share the record with your team, and monitor the upstream software your projects rely on.
Open-source CLI. macOS and Linux. No account needed to try.
Same reference
example/build-tool:stable
Different fingerprint. Review the change.
SCG compares the record, not just the name.
On your computer
Run the CLI from your local checkout.
Across your team
Commit one baseline for others to check.
In your build systems
Add the same check to automated workflows.
Make changes visible
A familiar name is not the whole story.
A tag can move while its name stays the same. SCG gives you a record to compare against, so a changed dependency becomes something you can investigate instead of something you simply inherit.
01
Record your baseline
SCG reads supported project files and records resolved dependency fingerprints in a signed scg.lock. Review that record before committing it to your repository.
scg init02
Check for drift
Compare the recorded entries with the platform’s latest available observations. SCG reports changed fingerprints and distinguishes an incomplete check from a finding.
scg check03
Keep watching
Register the lockfile with your organization to monitor its dependencies between local checks. Your plan determines how many repositories you can watch and how events are delivered.
scg watchOrganization sign-in required.
A change is a reason to investigate, not proof of an attack. A matching fingerprint is not proof that software is harmless.
Give third-party tools fewer opportunities to reach your credentials.
For GitHub Actions workflows, SCG’s secret audit compares environment-variable names with the tool’s security profile. Use the findings to review which credentials a step receives.
Results depend on the available profile and the environment where the audit runs.
Fits the tools you already use
One record across supported project dependencies.
Use the same CLI for supported package, container, and workflow references. Start from the files already in your repository.
JavaScript packages
npm registry dependencies from package-lock.json and pnpm-lock.yaml.
Python packages
Package references from requirements.txt and requirements-*.txt. Use exact version pins for a dependable baseline.
Container images
Base-image references from supported Dockerfiles, resolved through the platform’s Docker coverage.
GitHub Actions
Action references from .github/workflows/*.yml and .github/workflows/*.yaml.
Coverage depends on supported file syntax and indexed upstream references. SCG does not inventory every application installed on your computer.
Keep your lockfiles and vulnerability scanners.
SCG adds a cross-ecosystem baseline and upstream change history. It complements package-manager integrity checks, immutable references, and vulnerability scanning rather than replacing them.
Start with your own checkout
Local first. Ready for the rest of your workflow.
Install the CLI, create a baseline you have reviewed, and run your first check. You can add team monitoring or a build-system check afterward.
On your computer
macOS and Linux · Intel/AMD 64-bit and ARM64
Install SCG.
curl -sSfL https://scg.data-insights.ai/install.sh | shInspect the installerDownload a release instead
The installer checks the downloaded binary against the published checksum. Review the script before running it.
In your project root, record the baseline and review it.
scg init
cat scg.lockWhen you are satisfied with the record, commit it and run a check.
git add scg.lock
git commit -m "Add SCG dependency baseline"
scg checkCreating a baseline records what the platform resolves at that point. Review the dependencies you are choosing to trust; SCG does not certify that the first version you record is safe.
Local checks use the hosted SCG platform and require a network connection. Anonymous use has an hourly request allowance; an account is not required to begin.
Updating your baseline intentionally
Review the intended dependency changes first. Then regenerate the record and inspect the diff before committing it. Do not update the baseline merely to silence an unexpected drift finding.
scg update
git diff -- scg.lockWith your team
Share the record. Add ongoing monitoring.
Commit scg.lock so teammates can check the same baseline from their own computers. To register the repository for ongoing monitoring, create an organization and sign in from the CLI.
After creating and, where required, activating the organization, run:
scg login
scg watch
scg watches
scg intel --watchedThe Free plan watches one repository. Pro adds more watched repositories and signed webhooks. Review the plans and any organization verification charge before activating account-based API access.
In CI
Make the check an explicit gate.
Run SCG before the operations you intend to gate, after the repository and SCG are available. A nonzero result must not be treated as a successful verification.
GitHub Actions
Add this step after checkout in an existing Linux or macOS job. It runs the check against your committed scg.lock.
- name: Check dependency baseline
uses: data-insights-ai/rho-scg@343baa3c6f4ab29fc2b7bc89b7d6b7f2902574b1
with:
version: v0.4.0
mode: checkExample pinned to the v0.4.0 action commit and CLI release. Keep your checkout action and other dependencies pinned too. This check does not retroactively protect actions the runner has already fetched or executed.
Other build systems and scripts
Use the same CLI in GitLab, Jenkins, Buildkite, or your own scripts on a supported operating system. Read the exit status and the structured report rather than parsing colored terminal text.
scg check --json- 0 — Check completed: no drift was detected in the recorded entries.
- 1 — Finding or validation error: inspect the report and stop the gated operation.
- 2 — Check incomplete: investigate connectivity, limits, or stale data. Retry according to your policy; do not label this a detected attack or a successful check.
Shared upstream intelligence
Watch the dependencies your team actually uses.
Register your repositories’ lockfiles once, then use SCG’s watched-dependency feed to focus on those tools. You do not have to sift through every event in the software ecosystem to find the ones connected to your recorded dependencies.
When a reference changes, its history matters. SCG’s temporal knowledge graph keeps the observed relationships between references and fingerprints over time, so a change has context instead of being just another alert.
- Relevance: narrow events to watched dependencies.
- History: compare the previous record with later observations.
- Delivery: use signed webhooks on Pro and a private JSON/STIX feed on Enterprise.
Monitoring covers registered dependencies and available platform observations. Detection and delivery timing vary by source and plan; this is not continuous monitoring of processes on your computer.
The change has a history.
What changed? What was recorded before? Which watched dependencies match?
Take a first look
Check a public GitHub repository without installing anything.
Explore action references and declared secret exposure in public GitHub workflows. For checks against your project’s recorded dependency baseline, use the CLI.
This browser check covers public GitHub workflow files. It is not a malware scan or a full audit of the repository’s source code.
Know what you are relying on
A security tool should make its own boundaries clear.
Inspect the CLI, understand the data it sends, and see where the platform fits in the trust model.
Open-source CLI
The CLI is available under Apache 2.0. Review the implementation and published releases rather than relying only on a feature list.
Your baseline, signed
scg.lock is signed by the SCG platform and checked against the platform key built into the CLI. The signature protects the record’s integrity; it does not certify the dependencies as safe.
Clear data boundaries
Resolution and signing use the hosted platform. Lockfile metadata can include dependency references, fingerprints, repository identifiers, workflow paths, and secret-variable names—not secret values or application source code.
Built and operated by datAInsights GmbH in Vienna, Austria.
Start small. Expand when you need to.
Try the CLI. Add the monitoring your team needs.
Anonymous CLI access lets you get started without an account. Organization plans add watched repositories, history, and delivery options. A watched repository is not a limit on where you can install the CLI.
Free
€0
per month, after a one-off €1 against bulk signups
Watch your first repository. Opening the allowance costs €1 once: nothing to you, a great deal to somebody who wants ten thousand accounts, and filtering those out is the only job it has. It is not a price and it is not refunded. An account that has not paid it has no allowance of its own.
- Dependency resolution and baseline checks
- One watched repository
- Seven days of history
- One seat
- 100 dependency resolutions per hour per seat
- Email support without a guaranteed response time
For teams
Pro
€49
per seat / month
Or €490 per seat, billed yearly.
Monitor more repositories and bring changes into your workflow.
- Everything in Free
- Unlimited watched repositories
- Signed drift webhooks
- Blast-radius analysis
- 30 days of history
- 5,000 dependency resolutions per hour per seat
Enterprise
€499
per seat / month
Or €4,990 per seat, billed yearly.
Add organization-specific intelligence, profiles, and contracted support.
- Everything in Pro
- Private intelligence feed in JSON and STIX
- Custom organization security profiles
- No fixed history-retention limit in the plan
- 50,000 dependency resolutions per hour per seat
- Incident-response and support terms defined in your contract
A seat is a person with API access. Each seat has its own API keys and hourly allowance. Machine and repository counts are separate from seat counts.
The allowance counts dependency resolutions, not calls: a project with eight hundred recorded dependencies asks for eight hundred answers however few requests the CLI sends. That is why the numbers are what they are. Five Python requirements resolve to about nineteen packages and three or four npm packages to under a hundred, so Free covers a small project. React, React DOM, TanStack Query and Tailwind together are eighty entries in a lockfile; a real application is around a thousand dependencies and belongs on Pro; a monorepo is several thousand and belongs on Enterprise. Without a key you get 15 resolutions per hour per address, which covers the audit above and one to three packages by hand. There is no allowance for an account that has not paid the one-off €1. A quota given to anyone who can type an email address is a quota spent by scripts, and the people waiting behind them are the ones it was meant for. One euro sorts the two apart, which is all it is there to do.
Organization activation may require a one-time €1 verification payment, plus applicable tax. Any required payment is shown before you confirm. Anonymous CLI use does not require organization activation.
Prices are in euros and exclude VAT. Applicable tax is shown at checkout. Paddle is the merchant of record and handles payments and invoices.
Questions before you start?
Is SCG only for CI/CD?
No. You can run the CLI on your own macOS or Linux computer and check a local copy of your repository. The same baseline can be shared with teammates and checked in CI. SCG is project-based: it is not an antivirus or a background agent that monitors every process on your machine.
Does SCG replace my package lockfile or vulnerability scanner?
No. Keep package-manager lockfiles, integrity checking, immutable references, and vulnerability scanning. SCG adds a separate record across supported ecosystems and helps you track upstream changes to the dependencies you have recorded.
Does a passing check mean the dependency is safe?
No. It means no drift was detected for the recorded entries using the platform observations available to that check. Malicious software can be present in an initial baseline or an intentionally accepted update. Review what you trust and keep your other security controls.
Does SCG automatically block software on my computer?
No. The CLI reports a result. To gate a command or build, your script or build system must run the check and stop when verification does not succeed. The current check does not intercept every download or process, and it does not independently hash every installed file. See the security model for the precise boundary.
Do local checks work offline?
No. Dependency resolution and baseline signing use the SCG platform. An unavailable platform or stale data is not a successful check. Verification of a lockfile signature is a separate operation from verifying current dependency observations.
What information leaves my computer?
The CLI sends dependency references for resolution and lockfile metadata for signing. Watching a repository also uploads its signed lockfile. That metadata can include repository identifiers, workflow paths, step names, dependency fingerprints, and names of referenced secret variables. It does not include secret values or your application source code through these lockfile operations.
What is free, and when do I need an organization?
You can try the CLI anonymously within the per-IP hourly request allowance. An organization gives you account-based API access and lets you register watched repositories. The Free plan includes one watched repository; larger monitoring and delivery needs are covered by Pro and Enterprise. Check the pricing section for limits and any one-time organization verification payment.
Does every dependency change indicate an attack?
No. Some references are intentionally mutable, and legitimate releases can change them. SCG makes differences visible so you can investigate and decide whether to update your baseline. A drift event and a confirmed malicious package are different things.
Start with the dependencies in front of you.
Create a record you can inspect. Check it from your own computer. Bring the same baseline to your team when you are ready.