SOURCE CODE PROVENANCE • MULTI-LANGUAGE • PRIVACY-FIRST

How CodeSeal source provenance works

CodeSeal creates signed historical evidence about source code without turning your source repository into permanent cloud storage. It fingerprints supported source files, compares those fingerprints with prior CodeSeal observations, and records a signed provenance event.

Your code is processed. Your source is not the registry.

When you submit a ZIP archive or supported public repository, CodeSeal creates a temporary working copy, identifies supported source material, calculates cryptographic fingerprints, performs the requested comparison, and then retains the signed evidence needed for future provenance checks—not the submitted source tree itself.

01

Submit

Upload a source ZIP or use a supported public GitHub repository. Archive paths and extraction limits are checked before analysis.

02

Identify

CodeSeal separates supported source/configuration files from obvious caches, virtual environments, package stores, generated output, and build artifacts.

03

Fingerprint

SHA-256 file fingerprints, a package fingerprint, structural evidence, and PTCS provenance markers are calculated without executing submitted code.

04

Compare

Candidate fingerprints are compared with signed historical registry evidence to find exact files, relocated files, modified-known files, missing files, new files, and material overlap.

05

Sign & record

A registered observation receives stable PTCS identifiers and an Ed25519-signed event. Earlier observations remain intact; history is append-only in meaning.

06

Trace later activity

Future submissions can be checked against retained fingerprints. Matching evidence can create trace activity and notifications without needing the original source archive to remain stored.

LANG
LANGUAGE-AGNOSTIC REGISTRY

One provenance model across modern software stacks

CodeSeal is not a Python-only code scanner. Its core registry records source-tree evidence independently of programming language, while Phase 11 language-aware analysis identifies meaningful source blocks, reduces the weight of common constructions and obvious dependency/vendor material, and records attribution context without storing source text.

Python source-code support

Python

Python source, project metadata and virtual-environment exclusion.

Ruby source-code support

Ruby

Ruby source with Gem/Bundler-aware dependency context.

C source-code support

C

C source and headers with build-tree awareness.

C++ source-code support

C++

C++ source, headers and common native build metadata.

JavaScript source-code support

JavaScript

JavaScript modules and web application source.

TypeScript source-code support

TypeScript

TypeScript and TSX/modern module source.

SAP UI5 / OpenUI5 source-code support

SAP UI5 / OpenUI5

JavaScript/TypeScript, XML Views, fragments, manifests and UI5 project metadata.

HTML5 source-code support

HTML5

Web markup and application templates.

CSS / SCSS source-code support

CSS / SCSS

Stylesheets and supported preprocessors.

Java source-code support

Java

Java source with Maven/Gradle build context.

Kotlin source-code support

Kotlin

Kotlin source and Gradle-oriented project structure.

.NET / C# source-code support

.NET / C#

C#, F# and VB source with generated bin/obj separation planned.

Go source-code support

Go

Go source and module metadata.

Rust source-code support

Rust

Rust source and Cargo metadata with target output excluded.

ABAP / CDS source-code support

ABAP / CDS

ABAP source and Core Data Services definitions.

PHP source-code support

PHP

PHP application source and package metadata.

Swift / Objective-C source-code support

Swift / Objective-C

Apple-platform source formats and project code.

Scala source-code support

Scala

Scala source and JVM build context.

PowerShell / Shell source-code support

PowerShell / Shell

Automation scripts and shell source.

Vue source-code support

Vue

Vue single-file component source.

Svelte source-code support

Svelte

Svelte component source with generated build caches excluded.

GraphQL source-code support

GraphQL

GraphQL schema and query source.

Technology marks & privacy

The visual badges on this page are locally hosted CodeSeal technology identifiers rather than remote third-party tracking assets. Language, framework and product names remain the property/trademarks of their respective owners; their appearance here describes source formats CodeSeal can recognize and does not imply sponsorship or affiliation.

CodeSeal tries to measure authored source—not accidental baggage.

Common dependency and cache locations such as venv, .venv, site-packages, __pycache__, node_modules, package-manager caches, Rust target, framework build caches, dist, and other generated output are excluded from authored-source scanning where they can be identified safely.

Phase 11 already separates obvious dependency/vendor paths and low-uniqueness constructions from independently weighted authored-source blocks. Lockfiles, manifests, license headers and package metadata remain additional context for progressively stronger ecosystem classification rather than a reason to blindly ignore every folder named vendor.

A match does not automatically mean unattributed copying.

CodeSeal's Phase 11 provenance intelligence distinguishes raw technical similarity from attribution context. Proper source citations, public examples, license references, common helper functions, dependency/vendor material, and references to another PTCS event are classified separately from unexplained unique similarity.

Forensic reports now show both raw technical overlap and attribution-adjusted authored-source overlap. Verified PTCS references, declared citations, common constructions and dependency/vendor material remain visible as evidence but do not receive the same independent provenance weight as unexplained unique similarity.

Source code history for developers, teams, contractors and organizations

Traditional repository history is valuable, but source can move between repositories, archives, contractors, internal systems and deployment packages. CodeSeal adds a separate cryptographically signed observation layer that can help document when a source fingerprint was submitted and how later material relates to earlier recorded evidence.

That makes CodeSeal useful for software provenance checks, source code attribution review, historical code comparison, developer handoff evidence, contractor source history, internal software governance and forensic source matching—without claiming to replace copyright registration, contracts, repository history, software licenses or legal analysis.

READY TO CHECK A SOURCE TREE?

Scan first. Register only when you choose.

A no-history scan is free. Signing/registering source is free. CodeSeal only charges for detailed history inspection when history exists and the user chooses to inspect it.

Scan & search source →