SuperClaude-Org/SuperClaude_Framework

SuperClaude is a dual-platform (Python and Node.js) framework that provides a structured execution engine for AI agents within the Claude Code environment. It manages a library of specialized domain agents and enforces high-quality implementation through built-in confidence checks, parallel task execution, and self-correction mechanisms. The system distributes as a pytest plugin and CLI tool, integrating with external MCP servers to support complex, multi-step problem solving and workflow analysis.

A new 'Confidence Check' skill has been introduced to require a minimum 90% confidence score before starting any implementation. This skill enforces a pre-implementation assessment that verifies the absence of duplicate code, ensures architecture compliance with the existing tech stack, confirms official documentation has been reviewed, validates the existence of working open-source references, and ensures the root cause is clearly identified. The skill includes both a TypeScript implementation and a Markdown guide detailing the five weighted checks and the logic for proceeding, pausing, or stopping based on the calculated confidence level.

.claude, skills · high confidence

The scripts directory now includes a comprehensive set of tools for package distribution and project maintenance. Users can build and publish the SuperClaude package to PyPI (including TestPyPI) using the new \publish.sh\ wrapper and \build\_and\_upload.py\ script, with automated validation and installation testing. A new \build\_superclaude\_plugin.py\ script generates distribution artifacts for the SuperClaude plugin by rendering manifests and copying assets. Additionally, \ab\_test\_workflows.py\ and \analyze\_workflow\_metrics.py\ provide frameworks for comparing workflow variants and analyzing performance metrics (tokens, time, success rates) to identify inefficiencies and best practices. Maintenance is supported by \cleanup.sh\ for removing build artifacts and caches, and \uninstall\_legacy.sh\ for safely removing old SuperClaude configurations from the user's home directory.

scripts · high confidence

The SuperClaude package has been updated to version 4.3.0, introducing a comprehensive set of specialized AI agent definitions for the \src/superclaude/agents/\ directory. This release adds persona-based instructions for backend, frontend, and DevOps architecture, as well as dedicated agents for business strategy, deep research, performance engineering, quality assurance, security, and code refactoring. The package now exposes core \pm\_agent\ components (ConfidenceChecker, ReflexionPattern, SelfCheckProtocol) in its public API, providing users with a broader toolkit for structured, role-specific AI interactions.

src/superclaude · high confidence

The \setup/components\ directory has been removed, deleting the \CoreComponent\, \CommandsComponent\, \MCPComponent\, and \HooksComponent\ classes. This eliminates the previous modular installation logic that handled framework files, slash commands, MCP server integration, and hooks via individual component classes, marking a structural shift in how the SuperClaude setup process is organized.

setup/components · high confidence

The \setup/\_\init\\_.py\ file, which previously exposed the SuperClaude installation suite's version (3.0.0) and core path constants (such as \CONFIG\_DIR\ and \PROFILES\_DIR\), has been deleted. This change removes the legacy Python-based installation metadata and path definitions from the package namespace, aligning with the broader framework restructuring and removal of the old profile system.

setup · high confidence

The \setup/utils\ package has been removed, deleting the \\_\init\\_.py\, \logger.py\, \security.py\, and \ui.py\ files. This eliminates the previous installation-time logging system, path and filename security validation logic, and the custom console UI components (progress bars, menus, and colorized output) that were previously exposed through this module.

setup/utils · high confidence

The \setup/core\ package, which previously provided core installation utilities such as \ConfigManager\, \SettingsManager\, \FileManager\, \Validator\, and \ComponentRegistry\, has been completely removed. This change eliminates the legacy configuration loading, schema validation, cross-platform file operations, and component auto-discovery logic that were previously centralized in this directory, indicating a structural shift in how the SuperClaude installation system manages its setup dependencies and workflows.

setup/core, setup/operations · high confidence

The \setup/base\ module has been removed, deleting the \Component\ abstract base class and the \Installer\ orchestrator that previously handled dependency resolution, system validation, and backup creation for the installation system.

setup/base · high confidence

The project has been restructured to follow the standard Python src layout, moving package code into src/superclaude/ and tests into tests/ to prevent accidental imports. Development and runtime operations now exclusively use uv (e.g., uv run pytest, uv pip install), replacing legacy setup scripts and pip commands. A new .pre-commit-config.yaml enforces code quality and security by integrating detect-secrets, conventional commit validation, and custom checks to block hardcoded secrets. Additionally, a .env.example file has been added to document required PyPI API tokens and GitHub secrets for CI/CD.

(repo-wide) · high confidence

This release introduces a major framework restructuring (V4 Beta) that replaces the previous command definitions with new Python-based execution engines and a new synchronization script. The \scripts/sync\_from\_framework.py\ script is added to automate pulling content from the SuperClaude\_Framework repository and transforming it for distribution as a Claude Code plugin with namespace isolation (sc: prefix). The \src/superclaude/execution/\ directory now contains new Python modules: \parallel.py\ for automatic parallelization of independent tasks, \reflection.py\ for a 3-stage pre-execution confidence check, and \self\_correction.py\ for learning from past failures. Concurrently, the old \SuperClaude/Commands/\ markdown files (e.g., analyze.md, build.md, task.md) and \SuperClaude/Core/\ configuration files (e.g., COMMANDS.md, FLAGS.md) are removed, indicating a shift from static markdown-based command definitions to a more dynamic, code-driven execution model.

SuperClaude · high confidence

The SuperClaude plugin has been restored and updated to version 4.3.0, reinstating the full framework of 30 slash commands, 20 domain-specialist agents, and 7 skills. This update introduces configuration for two new Model Context Protocol (MCP) servers—Context7 for official library documentation and Sequential Thinking for multi-step problem solving—while also updating the installation directory to \~/.claude/commands/sc/.

plugins/superclaude · high confidence

Added a comprehensive test suite for the SuperClaude Framework, including unit tests for core components (CLI install commands, confidence checking, parallel execution, reflection, reflexion, self-check, and self-correction) and integration tests for the execution engine and pytest plugin. The suite introduces shared pytest fixtures in conftest.py to support testing confidence assessments, self-check protocols, and PM agent context structures.

tests · high confidence

This change introduces the official build manifests for SuperClaude v4.3.0, enabling distribution via both PyPI (Python) and NPM (Node.js). The Python package (\pyproject.toml\) defines the core framework as a pytest plugin with CLI capabilities, requiring Python 3.10+ and dependencies like pytest, click, and rich. The NPM wrapper (\package.json\) provides a Node.js interface for the same framework, targeting Node 16+ and including post-install scripts for setup. This establishes the foundational packaging structure for the dual-platform release.

(dependencies) · high confidence

Written by watchdog.canine.dev from the codebase's own history, inside the signed delivery this page is composed from.

• First survey — no prior run to compare against. CAI 69.

• Code Health 94
• Architecture 89
• Maturity 71
• Readiness 63
• Security 69

• 300 commits — 237 feature/other, 63 fixes

• (root) — 120 commits
• setup/components — 22 commits
• (repo) — 19 commits
• src/superclaude — 15 commits
• .github/workflows — 11 commits
• Docs/Reference — 7 commits
• Docs/User-Guide — 7 commits
• SuperClaude/Commands — 7 commits
• SuperClaude/main.py — 7 commits
• setup/cli — 5 commits
• setup/operations — 5 commits
• SuperClaude/Agents — 4 commits
• plugins/superclaude — 4 commits
• Docs/installation-guide.md — 3 commits
• Framework-Hooks/docs — 3 commits
• Guides/installation-guide.md — 3 commits
• SuperClaude/Hooks — 3 commits
• docs/Development — 3 commits
• setup/base — 3 commits
• setup/utils — 3 commits

• fix: Claude/fix readme plugin docs 011 cv2 lk n gko6wzm tfn dmnrm (#478)
• fix: Fix AIRIS MCP Gateway installer bugs (#525)
• fix: Fix MCP installer and package configurations
• fix: Fix MCP installer and package configurations (#338)
• fix: Fix MCPDocsComponent initialization error
• fix: Fix ModuleNotFoundError in PyPI publish workflow (#433)
• fix: Fix SuperClaude installation and backup issues
• fix: Fix component validation and bump version to 4.0.6 (#292)
• fix: Fix installation failures on Windows systems with alias usernames
• fix: Fix installation failures on Windows systems with alias usernames (#213)
• fix: Fix mcp_docs component initialization order issue
• fix: Fix troubleshoot command usage documentation (#220)
• fix: Fix update command and installer logic
• fix: Fix update command and installer logic (#339)
• fix: Fixed Some Bugs (#355)
• fix: Fixed installer.py
• fix: Fixed serena installation by add serena to claude directly (#357)
• fix: Fixes (#379)
• fix: Fixes (#384)
• fix: Fixes and improvement (#378)
• …and 280 more

• 0 containers · 1 bounded contexts · 0 dependency edges (baseline)

Written by watchdog.canine.dev from the codebase's own history, inside the signed delivery this page is composed from.

SuperClaude-Org/SuperClaude_Framework was measured the same way every project in this corpus was: the same rubric, at a pinned commit, with the result published in full. Point a surveyor at a repository you know and see whether you agree with it.