Skip to content

About

Cross-Host Autonomous Skill Router & Blocking Gates for Claude Code, Codex & Antigravity (158 Skills, 21 Auto-Routes)

Topics

Resources

Contributing

Security policy

Stars

0 stars

Watchers

0 watching

Forks

Latest commit

 

History

131 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

🌌 AgentLogic: Autonomous Evolving Rule Engine (V6.9.1)

Agentic Code License Version

AgentLogic is a versioned primary-logic package designed to be pulled into local AI coding hosts (Antigravity / Gemini, Claude Code, OpenAI Codex / GPT) and used as their default operating mind. It consolidates multi-turn session workflows, Turn 1 autonomous role recommendation, pure UI/UX design + text-to-image mockup coupling, zero-manual-selection domain skill routing, security sandbox auditing, and project-native knowledge synchronization.

Important

Repository Gating Warning: This repository represents core logic infrastructure, not a generic business workspace. Before starting any code modifications or host installations, AI agents MUST anchor to 00_START_HERE.md and complete the mandatory startup logic entry check.


🚀 Zero-Friction Install (Let Your Agent Do It)

You don't need to manually copy 158 skill folders. Just paste this repository to your coding agent (Antigravity, Claude Code, or OpenAI Codex):

Prompt for your Agent:
Read AGENT_INSTALL.md in this repository, install AgentLogic for my local host non-destructively, and verify with scripts/doctor.ps1.

Manual PowerShell Commands (Optional)
Step PowerShell Command Primary Purpose
1. Diagnosis powershell -NoProfile -ExecutionPolicy Bypass -File .\scripts\doctor.ps1 Analyzes repository layout, checks manifests, and recommends installers
2. Install powershell -NoProfile -ExecutionPolicy Bypass -File .\install.ps1 -TargetHost antigravity Validates, builds, verifies, and deploys the host-ready ruleset and 11 first-party skills (antigravity / claude / codex)
3. Validation powershell -NoProfile -ExecutionPolicy Bypass -File .\scripts\maintainer-check.ps1 Verifies structural link integrity, 158 registered skills, 21 auto-routes, and compiled adapter bundles

⚡ Why AgentLogic? (vs. Vanilla AI & Ad-hoc Skills)

Dimension Vanilla AI (Raw Prompting) Ad-hoc Skill Users AgentLogic (V6.9.1)
Skill Dispatch Relies on ad-hoc prompting; resets every new session Requires remembering & manually invoking skills Zero-Manual-Selection: Turn 1 role router + 21 deterministic auto-routes
Blocking Gates AI often skips planning and dumps boilerplate code Soft suggestions only; skills can be bypassed Hard Blocking Gates: 1-question requirements brief → ui-ux-pro-max spec → text-to-image mockup approval before UI code
Real-World Visuals & Motion Hallucinates anatomy/geometry for SVGs & animations No visual/motion reference enforcement Real-World Reference Grounding: Requires real photos/GIFs/short videos before coding SVGs, Canvas, or animations of real-world subjects
Modern Web & Extensions Frequently defaults to legacy CSS/JS workarounds Fragmented or outdated frontend guidance GoogleChrome/modern-web-guidance (209 Baseline guides + legacy-scan) & Chrome Extensions MV3 built into auto-routes
Cross-Host Parity Rules diverge across Claude, Codex, and Antigravity Separate configs per tool Single Source of Truth: Compiles to Antigravity, Claude Code, Codex, and Generic + PROJECT_MAP.md / CODE_MAP.md sync

🏛️ 4-Layer Architecture & V6.9.1 Autonomous Execution Flow

We organize rules, blocking gates, and auto-triggered capabilities into a 4-layer progressive abstraction model so every host executes the same deterministic workflow without manual /skill selection:

flowchart TD
    subgraph L1 ["L1: Strategy, Entry & Role Gate (Turn 1 Blocking)"]
        Entry["MAIN.md / AGENTS.md / Host Global Rules"] --> Gate1{"Gate 1: Startup & Context Gate"}
        Gate1 -->|"Ask forget/compress context"| RoleRouter["role-persona-router (Auto #1 Role Recommendation + 7-Role Menu)"]
        RoleRouter --> Check["00_START_HERE.md & Knowledge Sync (PROJECT_MAP / CODE_MAP)"]
    end

    subgraph L2 ["L2: Session, SDD & Automatic Task Router (21 Routes)"]
        Check --> Router{"Gate 3: Encounter -> Auto-Route"}
        Router -->|"UI / Frontend Task"| Grilling["product-requirements-grilling (1-Question-at-a-Time Brief)"]
        Router -->|"Small Fix / Modernization"| WebMod["web-platform-modernization (modern-web-guidance / legacy-scan)"]
        Router -->|"Chrome Extension"| ExtRoute["chrome-extension-development (chrome-extensions MV3)"]
        Router -->|"Feature / Refactor"| SDD["38_SpecDrivenDevelopment.md (Spec -> Red-Team -> Plan -> Tasks)"]
        Router -->|"Domain / Creative / Reverse Task"| DomainDispatch["Zero-Manual-Selection Domain Dispatch"]
    end

    subgraph L3 ["L3: Design, Reference Grounding & Verification Gates"]
        Grilling --> UIProMax["Gate 2: ui-ux-pro-max (Pure Palette / Typography / Grid Spec)"]
        UIProMax --> MockupGate["Mandatory Text-to-Image UI Mockup + User Approval Gate"]
        MockupGate --> RefGate["Real-World Visual & Motion Reference Grounding (Photos / GIFs / Video)"]
        RefGate --> UIImpl["modern-web-guidance + Frontend Implementation & Visual Verification"]
        SDD --> CodeSimp["code-simplifier (Post-Implementation AST & Guard-Clause Polish)"]
        CodeSimp --> Verify["29_MandatoryVerificationMatrix.md & SRA Audit"]
    end

    subgraph L4 ["L4: First-Party Auto-Triggered Skills & Drivers (158 Portable Skills)"]
        UIImpl --> BrowserSkill["modern-web-guidance, chrome-extensions, agent-browser & nano-banana-pro"]
        DomainDispatch --> Decompile["decompile-reconstruction (IDA / Ghidra / jadx / webcrack -> Clean-Room Source)"]
        DomainDispatch --> MediaDir["multimodal-media-director (Text-to-Image / Video Storyboard / Music Arrangement)"]
        DomainDispatch --> OfficePro["office-pro-suite (Executive Word .docx / Dynamic Excel .xlsx / Action-Title PPT .pptx)"]
        DomainDispatch --> NovelArch["novel-story-architect (.story-bible/ Continuity + Scene Beats + Anti-AI-Cliche)"]
        DomainDispatch --> DiagramArch["diagram-flowchart-architect (.drawio XML / .excalidraw JSON / Themed Mermaid)"]
    end

    classDef l1 fill:#dbeafe,stroke:#1d4ed8,color:#1e3a8a,stroke-width:2px
    classDef l2 fill:#fef3c7,stroke:#b45309,color:#78350f,stroke-width:2px
    classDef l3 fill:#fce7f3,stroke:#be185d,color:#831843,stroke-width:2px
    classDef l4 fill:#dcfce7,stroke:#15803d,color:#14532d,stroke-width:2px

    class Entry,Gate1,RoleRouter,Check l1
    class Router,Grilling,WebMod,ExtRoute,SDD,DomainDispatch l2
    class UIProMax,MockupGate,RefGate,UIImpl,CodeSimp,Verify l3
    class BrowserSkill,Decompile,MediaDir,OfficePro,NovelArch,DiagramArch l4
Loading

Detailed End-to-End Execution & Blocking Gate Workflow

flowchart TD
    Start(["User Request"]) --> ContextCheck["Gate 1a: Context-Reset Check (Continue or Forget/Compress?)"]
    ContextCheck --> RoleGate["Gate 1b: role-persona-router (Recommend #1 Lead Role + 7-Role Menu)"]
    RoleGate --> SyncCheck["Knowledge Sync (git fetch + AGENTS.md / PROJECT_MAP.md / CODE_MAP.md / .learnings/)"]
    SyncCheck --> Classify{"Deterministic Task Classification (21 Auto-Routes)"}

    %% Frontend Net-New UI Path
    Classify -->|"Net-New UI / Substantial Refactor"| ReqGrill["product-requirements-grilling (1 Question at a Time -> requirements-brief.md)"]
    ReqGrill --> DesignSpec["ui-ux-pro-max (Industry Palette, Typography, Grid Spec)"]
    DesignSpec --> MockupGen["Text-to-Image Mockup (image-generator / multimodal-media-director / nano-banana-pro)"]
    MockupGen --> MockupApprove{"User Approved or Explicit Opt-Out?"}
    MockupApprove -->|"Revise"| DesignSpec
    MockupApprove -->|"Approved"| RefCheck{"Real-World Object or Motion (SVG / Canvas / 3D / Anim)?"}
    RefCheck -->|"Real-World Subject"| FetchRef["Inspect Real Photos (Static) & GIFs / Video (Motion Keyframes)"]
    RefCheck -->|"Abstract / Fictional"| DirectCreate["Proceed to Creative Composition"]
    FetchRef --> WebGuide["modern-web-guidance (Baseline APIs, Fallbacks, CWV, A11y)"]
    DirectCreate --> WebGuide
    WebGuide --> VisualVerify["Browser & Bounding-Box Screenshot Verification"]

    %% Frontend Small Fix & Chrome Extension Paths
    Classify -->|"Small Frontend Fix / Legacy Modernization"| WebModRoute["web-platform-modernization (modern-web-guidance / legacy-scan)"]
    WebModRoute --> RefCheck
    Classify -->|"Chrome Extension (MV3)"| ExtDev["chrome-extension-development (chrome-extensions MV3, CSP Sandbox, Store Checklist)"]
    ExtDev --> VisualVerify

    %% Engineering & SDD Path
    Classify -->|"Feature / Refactor / Multi-File"| SDDGate["SDD Lifecycle (Spec -> Red-Team Gate -> Plan -> Tasks -> Implement)"]
    SDDGate --> Simplifier["code-simplifier (Lock Green -> Guard Clauses / ast-grep -> Re-Verify)"]
    Simplifier --> FinalVerify["Verification Matrix & Fresh Command Output"]

    %% Specialized Domain Auto-Dispatch
    Classify -->|"Binary / Bytecode / Obfuscated JS"| Decomp["decompile-reconstruction (IDA / Ghidra / jadx / ilspycmd / pycdc / webcrack)"]
    Classify -->|"Word / Excel / PPT"| Office["office-pro-suite"]
    Classify -->|"Image / Video / Music"| Media["multimodal-media-director"]
    Classify -->|"Novel / Script"| Novel["novel-story-architect"]
    Classify -->|"Flowchart / Architecture Diagram"| Diagram["diagram-flowchart-architect"]

    Decomp --> FinalVerify
    Office --> FinalVerify
    Media --> FinalVerify
    Novel --> FinalVerify
    Diagram --> FinalVerify
    VisualVerify --> FinalVerify

    FinalVerify --> MapSync["Update PROJECT_MAP.md / CODE_MAP.md / .learnings/ERRORS.md"]
    MapSync --> Done(["Verified Delivery"])

    classDef gate fill:#fee2e2,stroke:#b91c1c,color:#7f1d1d,stroke-width:2px
    classDef ui fill:#fce7f3,stroke:#be185d,color:#831843,stroke-width:1.5px
    classDef eng fill:#dbeafe,stroke:#1d4ed8,color:#1e3a8a,stroke-width:1.5px
    classDef domain fill:#dcfce7,stroke:#15803d,color:#14532d,stroke-width:1.5px

    class ContextCheck,RoleGate,MockupApprove,RefCheck gate
    class ReqGrill,DesignSpec,MockupGen,FetchRef,DirectCreate,WebGuide,WebModRoute,ExtDev,VisualVerify ui
    class SDDGate,Simplifier,FinalVerify,MapSync eng
    class Decomp,Office,Media,Novel,Diagram domain
Loading

Detailed Module Mapping

Abstraction Layer Primary Core Responsibility Included Files / Directories
L1: Strategy & Entry Defines entrypoints, Turn 1 context-reset confirmation, autonomous #1 role recommendation (role-persona-router), and project-native knowledge sync. - 00_START_HERE.md (起手式)
- MAIN.md (Stable Host Binding)
- AGENTS.md (Agent Entry)
- 15_LogicEntryChecklist.md (Gate Checklist)
- 54_V6_9_RoleAndMultimodalCreativeGovernance.md (V6.9 Role & Creative Gates)
L2: Session & Loop Manages the single-turn execution cycle, single-question requirements grilling, SDD lifecycle, and 21-route zero-manual-selection dispatch. - 03_AgenticLoop.md (Agentic Cycle)
- 38_SpecDrivenDevelopment.md (SDD Lifecycle)
- 48_V6_8_SkillAndToolRouting.md (Autonomous Skill & Tool Routing)
- 53_V6_8_ProgrammingLevelSkillEnforcement.md (Skill Enforcement State Machine)
L3: Control & Auditing Establishes pure UI/UX design-first + mandatory text-to-image mockup gating, post-implementation code simplification, and verification matrices. - 29_MandatoryVerificationMatrix.md (Verification Matrix)
- 39_TasteFrontendDesign.md (Aesthetic & Spacing Audit)
- 30_ToolQuestionBoundaryMatrix.md (Credential Boundary)
- 31_ToolMisuseAndAntiPatternMatrix.md (Anti-Pattern Gating)
L4: Base & Drivers 11 first-party auto-deployed domain skills across all hosts (~/.gemini/skills, ~/.claude/skills, ~/.codex/skills), 158 portable skills, and original UI templates. - skills/ (11 First-Party Auto-Triggered Skills + 158 Registered Portable Skills)
* role-persona-router (Turn 1 Role Recommendation & 7-Role Menu)
* product-requirements-grilling (Single-Question Requirements Brief)
* ui-ux-pro-max (Pure UI/UX Design Spec + Mandatory Text-to-Image Coupling)
* modern-web-guidance (Modern Web Platform APIs, Baseline Compatibility, CWV & legacy-scan)
* chrome-extensions (Manifest V3 Browser Extensions & Chrome Web Store Publishing)
* multimodal-media-director (Text-to-Image / Video Storyboard / Music Production)
* office-pro-suite (Executive Word .docx, Excel .xlsx, PowerPoint .pptx)
* novel-story-architect (.story-bible/ Continuity & Anti-AI-Cliché Fiction Engine)
* diagram-flowchart-architect (.drawio XML, .excalidraw JSON & Themed Mermaid)
* code-simplifier (Auto Post-Implementation Code Refinement & AST Rewrite)
* decompile-reconstruction (Auto Decompilation & Clean-Room Source Reconstruction)
- templates/frontend/ (Original, unbranded UI starting points)

🛠️ Self-Improvement & Learning Feed

This rule package supports automated self-evolution and developer logging:

flowchart LR
    A["Unexpected Error / User Correction / Best Practice"] -->|"1. Log to"| B[".learnings/ (LEARNINGS.md & ERRORS.md)"]
    B -->|"2. Recurrence >= 3"| C["Promote Systemic Rule into AGENTS.md / CLAUDE.md / CODE_MAP.md"]
Loading

Tip

Pure Design + Mandatory Text-to-Image Mockup Gate (ui-ux-pro-max): Net-new UI and substantial UI refactors MUST first complete single-question requirements grilling, derive a pure ui-ux-pro-max visual design specification (industry palette, font pairing, spatial grid), and generate a text-to-image mockup for user approval before any frontend code is written.

🎨 Original Frontend Template Library

templates/frontend/ contains a small, dependency-free set of original, unbranded HTML/CSS starting points: an editorial landing page, an analytics dashboard, and a settings surface. They complement ui-ux-pro-max and Taste governance rather than replacing them: ui-ux-pro-max governs pure visual design specification and mandatory text-to-image mockup synthesis, while the templates provide reusable structural baselines. They contain no third-party images, fonts, logos, remote assets, or copied Open Design code.


📚 External References & Integrated Sources

During its evolutionary cycles, AgentLogic has integrated concepts and structures from the following open-source GitHub projects:

Project Source GitHub Repository Primary Concept Integrated
Linux Kernel Map makelinux/linux_kernel_map Inspired the 4-layer structural system layout grid (40_AgentLogicSystemMap.md)
Agent Browser vercel-labs/agent-browser Provided deterministic ref-based headless browser automation optimized for AI agents
Nano Banana Pro google-gemini/imagen Supported high-fidelity visual UI mockup generation and layout prototyping for design-first workflows
Self-Improvement Agent peterskoett/self-improving-agent Enabled structured error logging, user-correction tracking, and recursive pattern hardening
Spec Kit github/spec-kit Formulated Spec-Driven Development (SDD) code-writing phases and task constraints
Taste Skill Leonxlnx/taste-skill Outlined physics spring animations, HSL palette controls, and typographical kerning guidelines
PM Skills (AI Shipping) phuryn/pm-skills Establishes the AI Shipping & Reviewability Audit (SRA) protocol to restore human reviewability
Ponytail (Lazy Dev) DietrichGebert/ponytail Enforces the 7-rung ladder (YAGNI, reuse, stdlib, native, dependency, one-line, min-code) to cut code bloat
Matt Pocock Skills mattpocock/skills Integrates interactive grilling, CONTEXT.md vocabulary maps, and Architecture Decision Records (ADRs)
Cybersecurity Skills Router zhaoxuya520/reverse-skill Outlines malware analysis, APK reversing, frontend JS decryption, and firmware audit routing
Grok Build (comparative reference) xai-org/grok-build Compares provider-neutral runtime execution, state-preserving compaction, and deny-overrides policies
Graphify (external code-map reference) Graphify-Labs/graphify Provides an optional queryable project graph and user-approved code-view preview workflow
MengTo Skills (workflow comparison) MengTo/Skills Informs optional reference-to-interface capture, extraction, and verification workflows without importing external skills
OfficeCLI (Office-document workflow comparison) iOfficeAI/OfficeCLI Defines a mandatory route for requested Word/Excel/PowerPoint processing once separately trusted and installed; no binary, skill, or runtime is imported
PixelRAG (visual-document workflow comparison) StarTrail-org/PixelRAG Defines an optional visual-document discovery policy; it does not replace browser or screenshot verification
Firecrawl (external-collection workflow comparison) firecrawl/firecrawl Defines an optional consent-bound public-web collection policy; no API, SDK, or service is bundled
img2threejs (interactive 3D-web workflow comparison) img2threejs/img2threejs Defines an optional reference-image-to-procedural-Three.js model route for approved interactive 3D web work; no skill or runtime is bundled
Reverse Engineer Anything (REA) morluto/rea Informs the automatic decompile-reconstruction clean-room binary/bundle source reconstruction workflow
GhidraMCP LaurieWired/GhidraMCP Provides the open-source headless binary decompilation MCP benchmark for decompile-reconstruction
Anthropic Official Claude Plugins (code-simplifier) anthropics/claude-plugins-official Informs the automatic post-implementation code-simplifier complexity reduction and guard-clause refinement pass
ast-grep ast-grep/ast-grep Provides structural AST search and multi-site refactoring rules inside code-simplifier
Superpowers obra/superpowers Establishes the test-verified Red-Green-Refactor verification loop for code refinement
UI/UX Pro Max Skill nextlevelbuilder/ui-ux-pro-max-skill Provides pure UI/UX design-system intelligence (palettes, font pairings, layout grids) coupled with text-to-image mockups
Modern Web Guidance GoogleChrome/modern-web-guidance Provides 209 modern web platform guides, Baseline compatibility, CWV/a11y rules, legacy-scan, and Chrome Extension MV3 best practices (Apache-2.0)
Google Cloud GenMedia Creative Studio GoogleCloudPlatform/vertex-ai-creative-studio Informs multimodal Text-to-Image (Imagen), Text-to-Video (Veo), and Text-to-Music (Lyria) orchestration
EachLabs Multimodal Skills eachlabs/skills Informs cross-model image, video storyboard, and audio generation workflows
MiniMax MCP MiniMax-AI/MiniMax-MCP Provides Hailuo Video and MiniMax Music MCP generation reference patterns
Claude Image Gen guinacio/claude-image-gen Informs aspect-ratio, reference-image guidance, and inpainting controls for UI and illustration mockups
Anthropic Official Document Skills anthropics/skills Establishes executive Word (.docx), dynamic-formula Excel (.xlsx), and action-title PowerPoint (.pptx) standards
The Agency (agency-agents) msitarzewski/agency-agents Informs the 7 specialist role personas in the Turn 1 Autonomous Role Recommendation Gate
BMAD-METHOD bmad-code-org/BMAD-METHOD Informs multi-role Lead + Supporting Squad orchestration and artifact handoffs
Oh-Story ClaudeCode zenstory-ai/oh-story-claudecode Informs web-novel pacing, Golden 3-Chapter structure, and anti-AI-cliche prose filtering
Story Skills danjdewhurst/story-skills Establishes the .story-bible/ continuity compiler and 4-point chapter audit
Awesome Novel Studio MJbae/awesome-novel-studio Informs the 5-stage macro-to-scene-beat narrative drafting pipeline
Draw.io Official MCP & Skill jgraph/drawio-mcp Provides native .drawio XML and orthogonal layout rules for enterprise flowcharts
Agents365 Draw.io Skill Agents365-ai/drawio-skill Informs multi-view architecture, swimlane, and ER schema generation in .drawio
Excalidraw Diagram Skill coleam00/excalidraw-diagram-skill Informs color-zoned hand-drawn system design diagrams in .excalidraw

⚖️ Maintenance Constraints

  1. Keep it Pure: No business logic or project-specific implementations should be written into AgentLogic rule files.
  2. Specify Before Writing: Always compile technical blueprints to .specify/spec.md and .specify/plan.md before touching code.
  3. Generate via Bundle: Never edit dist/ bundles manually. Always write updates to version files (e.g. versions/AgentLogic_V6_9_1.md), then run build-profile.ps1 to rebuild the adapter targets.

Important

Mandatory Version Update & Pitfall Checklist (逻辑版本更新必查红线): Whenever promoting a version or updating core rules/skills, you MUST check and synchronize all items below (derived from .learnings/ERRORS.md historical post-mortems):

  1. 11-File Version & Skill Sync: Synchronize manifest.json, versions/AgentLogic_<VER>.md, 00_START_HERE.md, AGENTS.md, MAIN.md, README.md, 00_README.md, AGENT_INSTALL.md, modules/01_MasterLogic.md (L2 + L4 9 first-party skills), modules/11_CrossDirectoryMap.md, modules/40_AgentLogicSystemMap.md, PROJECT_MAP.md, CODE_MAP.md, and CHANGELOG.md.
  2. CP1252 & UTF-8 Encoding Redlines: Keep all operational .ps1 scripts 100% pure ASCII without BOM (use \uXXXX escapes instead of literal — or CJK text); always pass -Encoding UTF8 to Get-Content; always write JSON/JS/MD using [System.Text.UTF8Encoding]::new($false) (no BOM, no \u9225 CP936 mojibake).
  3. PowerShell & Antigravity Pitfalls: Never use $host as a variable name; never use double-quoted @"..."@ for Markdown containing backticks (`f becomes \x0C); always inline Gates 1–4 in ~/.gemini/GEMINI.md (24 KB cap) and register ~/.gemini/skills.
  4. Skill YAML & Defender Defanging: Every SKILL.md must have YAML front-matter name:; security reference docs must defang raw <script> / onerror= strings to prevent Windows Defender false-positive quarantine.
  5. 4-Host Rebuild, Local Sync & Tag Alignment: Run scripts/maintainer-check.ps1 + scripts/audit-skill-duplicates.ps1 -FailOnCandidates, run install.ps1 across antigravity / claude / codex, and align the Git release tag with the final main commit.

V6.9.1 Role & Multimodal Creative Governance

The default core-enhanced profile includes modules/54_V6_9_RoleAndMultimodalCreativeGovernance.md alongside modules/44_V6_8_GrokBuildOperationalIntegration.md, modules/45_V6_8_GraphifyCodeMapIntegration.md, modules/47_V6_8_ReferenceToInterfaceWorkflow.md, modules/48_V6_8_SkillAndToolRouting.md, modules/52_V6_8_Img2ThreeJs3DWebIntegration.md, and modules/53_V6_8_ProgrammingLevelSkillEnforcement.md.

V6.9.1 Project-Native Knowledge Sync & Skill Enforcement

Business projects synchronize their own concise operational facts and symbol-level navigation with code through tracked AGENTS.md, PROJECT_MAP.md, CODE_MAP.md, and .learnings/ERRORS.md. See Project-Native Knowledge Sync, CHANGELOG.md, and Module 53.


🛡️ Security, Clean-Room & Third-Party Attribution Disclaimer

  1. Authorized Security & Clean-Room Research Only: The reverse-engineering, binary triage (decompile-reconstruction), and security auditing routing rules (skills/reverse-skill/) included in this repository are strictly intended for authorized security assessments, malware defense analysis, CTF competitions, educational research, and lawful clean-room interoperability reconstruction. Users must obtain explicit legal authorization before analyzing any target binary, application, or system.
  2. Third-Party Open-Source Acknowledgments: AgentLogic is a curated meta-governance framework that synthesizes methodologies and portable skill definitions from the open-source community (listed in the External References & Integrated Sources table above). All vendored or adapted skill packages (spec-kit, taste-skill, agent-browser, reverse-skill, modern-web-guidance, etc.) retain their original open-source licenses (MIT / Apache-2.0) and copyright notices within their respective subdirectories.
  3. No Proprietary Runtime Bundling: This repository contains only plain-text Markdown governance rules, JSON routing contracts, and PowerShell/Node validation scripts. It does not bundle any proprietary commercial CLI binaries or closed-source source code.

About

Cross-Host Autonomous Skill Router & Blocking Gates for Claude Code, Codex & Antigravity (158 Skills, 21 Auto-Routes)

Topics

Resources

Contributing

Security policy

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages