{"passport":{"unfragile":{"@version":"1.0","version":"2026-05","artifact":{"id":"pypi_pypi-opencv-python","slug":"pypi-opencv-python","name":"opencv-python","type":"repo","url":"https://github.com/opencv/opencv-python","page_url":"https://unfragile.ai/pypi-opencv-python","categories":["frameworks-sdks"],"tags":[],"pricing":{"model":"open_source","free":true,"starting_price":null},"status":"active","verified":false},"capabilities":[{"id":"pypi_pypi-opencv-python__cap_0","uri":"capability://automation.workflow.multi.platform.pre.built.wheel.distribution.for.opencv.bindings","name":"multi-platform pre-built wheel distribution for opencv bindings","description":"Distributes pre-compiled OpenCV Python bindings as binary wheels across Windows (x86_64, x86), Linux (x86_64, i686, aarch64), and macOS (x86_64, arm64) for Python 3.7-3.12, eliminating the need for developers to compile OpenCV from source. Uses manylinux2014 standard for Linux wheels to ensure compatibility across diverse distributions. The build system automatically manages git submodules to pull the correct OpenCV C++ source version, compiles it with platform-specific optimizations, and packages the resulting extension modules into distributable wheels via a custom pyproject.toml backend.","intents":["Install OpenCV in a Python environment without building from source","Deploy OpenCV to cloud/container environments with minimal dependencies","Support multiple Python versions and architectures from a single package","Avoid compilation toolchain setup and long build times on developer machines"],"best_for":["Data scientists and ML engineers prototyping computer vision applications","DevOps teams deploying OpenCV to Docker containers and cloud platforms","Developers targeting multiple platforms who need consistent binary behavior"],"limitations":["Pre-built wheels may not include custom OpenCV compilation flags or optional dependencies not included in the standard build","Binary compatibility limited to officially supported platforms and architectures; custom hardware or exotic Linux distributions may require source compilation","Wheel size varies significantly between headless and GUI variants (headless ~30-40% smaller due to no X11/Qt dependencies)"],"requires":["Python 3.7 or higher","pip or compatible package manager","Platform matching one of: Windows (x86_64/x86), Linux (x86_64/i686/aarch64), macOS (x86_64/arm64)"],"input_types":["Python package specification (e.g., pip install opencv-python)"],"output_types":["Installed Python extension module (cv2)","Binary wheel file (.whl)"],"categories":["automation-workflow","package-distribution"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"pypi_pypi-opencv-python__cap_1","uri":"capability://automation.workflow.package.variant.selection.for.environment.specific.optimization","name":"package variant selection for environment-specific optimization","description":"Provides four distinct package variants (opencv-python, opencv-contrib-python, opencv-python-headless, opencv-contrib-python-headless) allowing developers to select the exact feature set and dependencies needed for their deployment context. The build system conditionally includes GUI components (Qt 5, X11 libraries) only in non-headless variants, and conditionally includes contrib modules (extra algorithms, experimental features) only in -contrib variants. This is managed through CMake configuration flags passed during the C++ compilation stage, reducing final wheel size and dependency footprint by 30-40% for headless deployments.","intents":["Deploy OpenCV to headless servers or Docker containers without unnecessary GUI dependencies","Access experimental and contrib algorithms (face recognition, ML modules) beyond core functionality","Minimize wheel size and installation footprint for resource-constrained environments","Avoid pulling in heavy transitive dependencies like Qt or X11 libraries in server contexts"],"best_for":["Backend engineers deploying computer vision services in containerized/serverless environments","Researchers needing access to contrib modules for advanced algorithms","Teams with strict dependency management policies or minimal container images"],"limitations":["opencv-contrib-python includes experimental/unstable algorithms that may change between versions","Headless variants cannot display images via cv2.imshow() or use GUI-dependent features; requires alternative display mechanisms","Switching between variants requires uninstalling and reinstalling; no single installation supports all variants simultaneously"],"requires":["Python 3.7+","pip or compatible package manager","Knowledge of which variant matches deployment context (desktop vs server, core vs contrib)"],"input_types":["Package name selection (opencv-python, opencv-contrib-python, etc.)"],"output_types":["Installed Python module with variant-specific features","Variant-specific binary wheel"],"categories":["automation-workflow","package-distribution"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"pypi_pypi-opencv-python__cap_10","uri":"capability://automation.workflow.python.version.specific.wheel.building.for.3.7.3.12.compatibility","name":"python version-specific wheel building for 3.7-3.12 compatibility","description":"Builds separate wheels for each supported Python version (3.7, 3.8, 3.9, 3.10, 3.11, 3.12) by compiling against version-specific Python headers and ABI (Application Binary Interface). The build system detects the Python version during compilation and generates extension modules with the correct ABI tag (e.g., cp39 for CPython 3.9). The CI/CD pipeline maintains separate build jobs for each Python version, ensuring that wheels are compatible with their target Python interpreter. This approach avoids ABI incompatibilities that would occur if a single wheel were used across multiple Python versions, which have different memory layouts and API signatures.","intents":["Install OpenCV wheels compatible with specific Python versions without ABI conflicts","Support multiple Python versions in the same environment (e.g., via virtual environments)","Ensure binary compatibility between wheel and Python interpreter","Enable gradual migration between Python versions without breaking existing deployments"],"best_for":["Teams managing multiple Python versions across different projects or environments","Organizations gradually upgrading from older Python versions (3.7, 3.8) to newer ones","CI/CD systems testing against multiple Python versions"],"limitations":["Building for all Python versions increases CI/CD complexity and build time significantly (6+ separate jobs)","Python 3.7 and 3.8 are approaching end-of-life; support may be dropped in future releases","ABI incompatibilities mean wheels are not interchangeable; installing wrong version causes ImportError"],"requires":["Python 3.7 or higher","pip or compatible package manager that selects correct wheel based on Python version","Matching Python version between build and runtime environments"],"input_types":["Target Python version (3.7-3.12)","Python-version-specific headers and ABI information"],"output_types":["Python-version-specific wheel files with ABI tags (e.g., cp39-cp39-linux_x86_64.whl)","Version-compatible extension modules"],"categories":["automation-workflow","code-generation-editing"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"pypi_pypi-opencv-python__cap_2","uri":"capability://automation.workflow.semantic.versioning.with.package.revision.tracking","name":"semantic versioning with package revision tracking","description":"Implements a four-part versioning scheme (cv_major.cv_minor.cv_revision.package_revision) that decouples OpenCV upstream versions from packaging-specific revisions. For example, 4.8.0.2 indicates OpenCV 4.8.0 with package revision 2 (second packaging iteration). Development builds append git hash identifiers (e.g., 4.8.0+abc1234) to enable tracking of unreleased builds. This allows maintainers to issue packaging fixes, dependency updates, or platform-specific patches without waiting for upstream OpenCV releases, while maintaining clear traceability to the underlying OpenCV version.","intents":["Track which OpenCV version is installed alongside packaging-specific fixes","Distinguish between upstream OpenCV updates and packaging/distribution improvements","Identify development/pre-release builds by their git commit hash","Manage dependency updates and security patches independently from OpenCV releases"],"best_for":["DevOps teams managing OpenCV versions across multiple environments","Maintainers issuing packaging-only patches without upstream OpenCV changes","Developers debugging version-specific issues and needing precise build identification"],"limitations":["Package revision increments are manual; no automatic detection of when a new revision is needed","Development version identifiers (git hashes) are local to the build environment and not standardized across CI/CD systems","Semantic versioning does not communicate feature/API changes; requires reading OpenCV release notes separately"],"requires":["Understanding of four-part version scheme (cv_major.cv_minor.cv_revision.package_revision)","Access to version metadata via pip show or __version__ attribute"],"input_types":["Version string from package metadata"],"output_types":["Parsed version tuple (OpenCV version, package revision)","Git commit hash for development builds"],"categories":["automation-workflow","package-distribution"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"pypi_pypi-opencv-python__cap_3","uri":"capability://automation.workflow.custom.pep.517.build.backend.with.cmake.orchestration","name":"custom pep 517 build backend with cmake orchestration","description":"Implements a custom Python build backend (in _build_backend/backend.py) that extends PEP 517 standards to orchestrate complex multi-stage compilation. The backend manages git submodule initialization, invokes CMake to configure the OpenCV C++ build with platform-specific flags, compiles extension modules via setuptools, and packages results into wheels. This abstraction allows setup.py to remain relatively simple while delegating platform-specific logic (Windows MSVC vs Unix Makefiles, architecture detection, dependency resolution) to the backend. The backend also handles patching of OpenCV source code before compilation, enabling fixes or customizations without forking the upstream repository.","intents":["Compile OpenCV C++ code into Python extension modules with platform-specific optimizations","Manage complex multi-stage build workflows (submodule init → CMake configure → compile → package) from a single entry point","Apply patches to OpenCV source before compilation without maintaining a fork","Support multiple platforms and architectures with a single build script"],"best_for":["Maintainers building OpenCV wheels for distribution across multiple platforms","Developers customizing OpenCV compilation flags or applying patches","CI/CD systems requiring reproducible, platform-agnostic build definitions"],"limitations":["Custom backend adds complexity; debugging build failures requires understanding both PEP 517 and CMake","Platform-specific logic is scattered across setup.py, backend.py, and CMakeLists.txt, making it harder to understand the full build flow","Patching system requires manual maintenance; patches may break with upstream OpenCV updates"],"requires":["Python 3.7+","CMake 3.10+","C++ compiler (MSVC on Windows, GCC/Clang on Unix)","Git (for submodule management)","setuptools and wheel packages"],"input_types":["setup.py configuration","CMakeLists.txt from OpenCV submodule","Platform-specific environment variables"],"output_types":["Compiled Python extension module (cv2.so or cv2.pyd)","Binary wheel file (.whl)"],"categories":["automation-workflow","code-generation-editing"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"pypi_pypi-opencv-python__cap_4","uri":"capability://automation.workflow.git.submodule.based.opencv.source.pinning","name":"git submodule-based opencv source pinning","description":"Maintains the OpenCV C++ source code as a git submodule, allowing the opencv-python repository to pin exact OpenCV versions without duplicating source code. When a new OpenCV release is made, maintainers manually update the submodule reference to point to the new tag/commit. The build system automatically initializes and checks out the submodule during the build process, ensuring reproducible builds across different machines and CI/CD environments. This approach decouples the opencv-python release cycle from OpenCV's, allowing packaging updates or patches to be released independently.","intents":["Pin exact OpenCV version to ensure reproducible builds across environments","Update OpenCV version independently from packaging infrastructure changes","Avoid duplicating OpenCV source code in the opencv-python repository","Enable developers to build against specific OpenCV versions without manual source management"],"best_for":["Maintainers managing OpenCV releases and packaging updates","Developers building opencv-python from source who need reproducible builds","CI/CD systems requiring deterministic build inputs"],"limitations":["Submodule updates are manual; no automatic detection of new OpenCV releases","Developers unfamiliar with git submodules may encounter confusing state (detached HEAD, missing source files)","Cloning the repository requires --recursive flag; shallow clones may not include submodule history"],"requires":["Git 2.13+ (for submodule support)","Network access to clone both opencv-python and opencv repositories","Understanding of git submodule workflows"],"input_types":["OpenCV git tag or commit hash"],"output_types":["Checked-out OpenCV source code in submodule directory","Submodule reference in .gitmodules"],"categories":["automation-workflow","tool-use-integration"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"pypi_pypi-opencv-python__cap_5","uri":"capability://automation.workflow.multi.platform.ci.cd.matrix.builds.with.automated.testing","name":"multi-platform ci/cd matrix builds with automated testing","description":"Implements an automated CI/CD pipeline (documented in CI/CD Pipeline section) that builds wheels across 3 major operating systems (Windows, macOS, Linux), 5+ CPU architectures (x86_64, x86, i686, aarch64, arm64), and 6 Python versions (3.7-3.12) in parallel. Each platform has a dedicated workflow (Linux Build Workflow, macOS Build Workflow, Windows Build Workflow) that handles platform-specific build steps, compiler selection, and dependency installation. The pipeline automatically triggers on new OpenCV releases (via submodule updates) and packaging changes, runs basic smoke tests on built wheels, and publishes successful builds to PyPI. Docker build environments ensure consistent compilation across different CI/CD runners.","intents":["Automatically build and test wheels across all supported platforms and Python versions","Detect platform-specific build failures early before publishing to PyPI","Publish new releases to PyPI without manual intervention","Maintain consistent build environments across different CI/CD systems via Docker"],"best_for":["Maintainers releasing new OpenCV versions to PyPI","Teams requiring automated, reproducible builds across multiple platforms","Projects needing to validate binary compatibility before distribution"],"limitations":["CI/CD pipeline is tightly coupled to GitHub Actions; migrating to other CI systems requires rewriting workflows","Build times are long (30+ minutes per platform) due to C++ compilation; not suitable for rapid iteration","Smoke tests are basic; comprehensive testing requires manual validation or additional test infrastructure"],"requires":["GitHub Actions (or equivalent CI/CD system)","Docker for consistent build environments","PyPI credentials for publishing","Access to multiple platform runners (or cloud-based multi-platform support)"],"input_types":["Git commits/tags triggering builds","Platform and Python version matrix configuration"],"output_types":["Built wheel files (.whl) for each platform/Python combination","Build logs and test results","Published packages on PyPI"],"categories":["automation-workflow","tool-use-integration"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"pypi_pypi-opencv-python__cap_6","uri":"capability://automation.workflow.headless.variant.compilation.with.conditional.gui.dependency.exclusion","name":"headless variant compilation with conditional gui dependency exclusion","description":"Builds headless package variants by passing CMake flags that exclude GUI-related dependencies (Qt 5, X11 libraries, GTK) during the C++ compilation stage. The build backend detects the target variant (via environment variables or setup.py configuration) and conditionally invokes CMake with -DBUILD_SHARED_LIBS=OFF, -DWITH_QT=OFF, and similar flags. This results in significantly smaller wheels (30-40% reduction) and eliminates transitive dependencies on heavy GUI libraries, making headless variants ideal for Docker containers, serverless functions, and cloud deployments where GUI functionality is unnecessary. The Python bindings remain functionally identical except for GUI-specific functions (cv2.imshow, cv2.waitKey) which raise NotImplementedError.","intents":["Deploy OpenCV to server/container environments without GUI dependencies","Reduce wheel size and installation footprint for resource-constrained deployments","Avoid pulling in heavy transitive dependencies (Qt, X11) in headless contexts","Maintain a single codebase while supporting both desktop and server deployments"],"best_for":["Backend engineers deploying computer vision services in Docker/Kubernetes","Serverless/FaaS deployments with strict package size limits","Teams running OpenCV in cloud environments without display capabilities"],"limitations":["GUI functions (cv2.imshow, cv2.waitKey, cv2.namedWindow) are unavailable in headless variants; code must use alternative display mechanisms","Debugging visual issues is harder without GUI display capabilities; requires saving images to disk or streaming to external viewers","Some OpenCV features that depend on GUI libraries may be unavailable or have reduced functionality"],"requires":["Python 3.7+","Deployment context without GUI requirements (server, container, serverless)","Code that doesn't rely on cv2.imshow() or other GUI functions"],"input_types":["Package variant selection (opencv-python-headless or opencv-contrib-python-headless)","CMake configuration flags"],"output_types":["Compiled Python extension module without GUI dependencies","Smaller binary wheel file"],"categories":["automation-workflow","code-generation-editing"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"pypi_pypi-opencv-python__cap_7","uri":"capability://automation.workflow.manylinux2014.wheel.compatibility.standard.for.broad.linux.distribution.support","name":"manylinux2014 wheel compatibility standard for broad linux distribution support","description":"Builds Linux wheels using the manylinux2014 standard (PEP 599), which targets glibc 2.17 and older, ensuring compatibility across CentOS 7, Ubuntu 14.04+, Debian 8+, and most modern Linux distributions. The build system uses Docker containers with manylinux2014 base images to compile against the correct glibc version and system libraries, then uses auditwheel to verify and repair wheels by bundling necessary shared libraries (libstdc++, libgomp) into the wheel. This approach eliminates the need for users to have specific system libraries installed, making wheels portable across diverse Linux environments without requiring source compilation.","intents":["Distribute OpenCV wheels that work across diverse Linux distributions without source compilation","Ensure binary compatibility with older Linux systems (CentOS 7, Ubuntu 14.04+)","Avoid requiring users to install system-level dependencies (glibc, libstdc++, etc.)","Simplify deployment to Linux servers and containers with minimal setup"],"best_for":["Teams deploying OpenCV to heterogeneous Linux environments (multiple distributions/versions)","Users on older Linux systems (CentOS 7, Ubuntu 14.04) who cannot upgrade","Container/cloud deployments requiring broad Linux compatibility"],"limitations":["Manylinux2014 wheels are limited to glibc 2.17+; systems with older glibc (e.g., CentOS 6) are not supported","Bundled shared libraries increase wheel size by 10-20% compared to system-provided libraries","Some system-specific optimizations (e.g., CPU-specific SIMD flags) may not be available due to glibc compatibility constraints"],"requires":["Linux system with glibc 2.17 or newer","Python 3.7+","pip or compatible package manager"],"input_types":["Linux platform specification","manylinux2014 Docker base image"],"output_types":["Manylinux2014-compliant wheel file (.whl)","Bundled shared libraries within wheel"],"categories":["automation-workflow","package-distribution"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"pypi_pypi-opencv-python__cap_8","uri":"capability://safety.moderation.third.party.binary.license.management.and.disclosure","name":"third-party binary license management and disclosure","description":"Manages licensing for bundled third-party binaries (FFmpeg, Qt 5, and others) by including a LICENSE-3RD-PARTY.txt file in distributions that lists all licenses and their terms. FFmpeg is included in all wheels under LGPLv2.1, Qt 5 is included in non-headless Linux wheels under LGPLv3, and other binaries have their respective licenses documented. The build system automatically bundles these binaries during compilation and ensures license files are included in the final wheel. This approach provides legal transparency and ensures compliance with open-source licensing requirements, particularly important for commercial users who need to understand licensing obligations.","intents":["Understand licensing obligations for bundled third-party binaries","Ensure commercial use compliance with open-source licenses (LGPL, Apache 2.0)","Verify that dependencies don't conflict with project licensing requirements","Maintain legal transparency in distributed packages"],"best_for":["Commercial teams using OpenCV who need to understand licensing obligations","Legal/compliance teams reviewing open-source dependencies","Projects with strict licensing requirements (e.g., GPL-incompatible projects)"],"limitations":["License files are static; changes to bundled binaries require manual license file updates","LGPL licensing (FFmpeg, Qt) requires providing source code or object files to users; wheels alone may not satisfy LGPL requirements for all use cases","License compliance is the responsibility of the user; the package provides information but not legal advice"],"requires":["Understanding of open-source licenses (Apache 2.0, LGPL, etc.)","Access to LICENSE-3RD-PARTY.txt in the installed package"],"input_types":["Installed opencv-python package"],"output_types":["LICENSE-3RD-PARTY.txt file with third-party license information","License metadata in package metadata"],"categories":["safety-moderation","package-distribution"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"pypi_pypi-opencv-python__cap_9","uri":"capability://automation.workflow.architecture.specific.wheel.building.for.x86.x86.64.i686.aarch64.and.arm64","name":"architecture-specific wheel building for x86, x86_64, i686, aarch64, and arm64","description":"Builds separate wheels for each supported CPU architecture (x86, x86_64, i686, aarch64, arm64) by detecting the target architecture during the build process and passing architecture-specific compiler flags to CMake. The CI/CD pipeline maintains separate build jobs for each architecture, with aarch64 and arm64 builds running on appropriate runners (e.g., AWS Graviton instances for aarch64, Apple Silicon for arm64). This enables users to install architecture-native wheels that are optimized for their specific CPU, improving performance through architecture-specific SIMD instructions and memory layouts. The build system automatically selects the correct wheel during pip install based on the runtime Python interpreter's architecture.","intents":["Install architecture-optimized OpenCV wheels for better performance on specific CPUs","Support ARM-based systems (Raspberry Pi, Apple Silicon, AWS Graviton) with native wheels","Avoid cross-architecture emulation (e.g., running x86 code on ARM) which degrades performance","Enable deployment to diverse hardware platforms (servers, edge devices, mobile)"],"best_for":["Teams deploying OpenCV to ARM-based systems (Raspberry Pi, Apple Silicon, cloud ARM instances)","Performance-critical applications requiring architecture-optimized binaries","Edge computing deployments with diverse hardware platforms"],"limitations":["Building for all architectures requires access to multiple hardware platforms or emulation; CI/CD complexity increases significantly","i686 (32-bit x86) wheels are increasingly rare; many modern systems are 64-bit only","Architecture detection during pip install relies on correct Python interpreter architecture; mismatches can cause runtime errors"],"requires":["Python interpreter matching the target architecture (e.g., ARM64 Python for arm64 wheels)","pip or compatible package manager that supports architecture-specific wheel selection","Hardware or CI/CD runners supporting the target architecture"],"input_types":["Target architecture specification (x86, x86_64, i686, aarch64, arm64)","Architecture-specific compiler flags"],"output_types":["Architecture-specific wheel files (.whl) with architecture tags","Optimized binaries for target CPU"],"categories":["automation-workflow","code-generation-editing"],"confidence":0.5,"matches":0,"success_rate":0}],"trust":{"score":24,"verified":false,"data_access_risk":"high","permissions":["Python 3.7 or higher","pip or compatible package manager","Platform matching one of: Windows (x86_64/x86), Linux (x86_64/i686/aarch64), macOS (x86_64/arm64)","Python 3.7+","Knowledge of which variant matches deployment context (desktop vs server, core vs contrib)","pip or compatible package manager that selects correct wheel based on Python version","Matching Python version between build and runtime environments","Understanding of four-part version scheme (cv_major.cv_minor.cv_revision.package_revision)","Access to version metadata via pip show or __version__ attribute","CMake 3.10+"],"failure_modes":["Pre-built wheels may not include custom OpenCV compilation flags or optional dependencies not included in the standard build","Binary compatibility limited to officially supported platforms and architectures; custom hardware or exotic Linux distributions may require source compilation","Wheel size varies significantly between headless and GUI variants (headless ~30-40% smaller due to no X11/Qt dependencies)","opencv-contrib-python includes experimental/unstable algorithms that may change between versions","Headless variants cannot display images via cv2.imshow() or use GUI-dependent features; requires alternative display mechanisms","Switching between variants requires uninstalling and reinstalling; no single installation supports all variants simultaneously","Building for all Python versions increases CI/CD complexity and build time significantly (6+ separate jobs)","Python 3.7 and 3.8 are approaching end-of-life; support may be dropped in future releases","ABI incompatibilities mean wheels are not interchangeable; installing wrong version causes ImportError","Package revision increments are manual; no automatic detection of when a new revision is needed","builder identity is not verified yet","no observed match outcomes yet"],"rank_breakdown":{"adoption":0.05,"quality":0.32,"ecosystem":0.3,"match_graph":0.25,"freshness":0.9,"weights":{"adoption":0.3,"quality":0.2,"ecosystem":0.15,"match_graph":0.3,"freshness":0.05}},"observed_outcomes":{"matches":0,"success_rate":0,"avg_confidence":0,"top_intents":[],"last_matched_at":null},"maintenance":{"status":"active","updated_at":"2026-05-24T12:16:25.060Z","last_scraped_at":"2026-05-03T15:20:20.420Z","last_commit":null},"community":{"stars":null,"forks":null,"weekly_downloads":null,"model_downloads":null,"model_likes":null}},"distribution":{"claim_url":"https://unfragile.ai/submit?claim=pypi-opencv-python","compare_url":"https://unfragile.ai/compare?artifact=pypi-opencv-python"}},"signature":"NkTPPAEChVPnlYWZzCnzUjfnELUH2OyB1R9E8qsjt7xd7wre0bdjB7i7aV098+Xz7f+H5ErxnlMDdo1GVT5YCg==","signedAt":"2026-06-16T05:24:30.460Z","signedBy":"unfragile.ai","version":1},"_links":{"self":"https://unfragile.ai/api/v1/passport/pypi-opencv-python","artifact":"https://unfragile.ai/pypi-opencv-python","verify":"https://unfragile.ai/api/v1/verify?slug=pypi-opencv-python","publicKey":"https://unfragile.ai/api/v1/trust-passport-public-key","spec":"https://unfragile.ai/trust","schema":"https://unfragile.ai/schema.json","docs":"https://unfragile.ai/docs"}}