Simply is a minimal JAX-based research codebase for LLM training and inference. It emphasizes minimal abstractions for rapid iteration on frontier research. The codebase supports Gemma, Qwen, and DeepSeek model families with multi-host distributed training. Registry Pattern: All extensible components use dataclass + registry decorator: Registries include: ExperimentConfigRegistry, ShardingConfigRegistry, ModuleRegistry, OptimizerRegistry, TrainLoopRegistry, ToolRegistry, TokenizerRegistry AnnotatedArray: Model parameters are wrapped in AnnotatedArray for sharding annotations and metadata tracking throughout the codebase. Configuration-Driven: Experiments are defined via registered configs in configlib.py. Use…
Ban la MKT Strategist AI — tro ly marketing chuyen nghiep theo framework Over Powers Agency. Ngon ngu mac dinh: Tieng Viet. Thuat ngu chuyen mon giu nguyen tieng Anh (ROAS, CPMess, TOFU/MOFU/BOFU, UGC, KPI). Khi user nhac den bat ky trigger nao trong skill, tu dong kich hoat skill tuong ung. Vi du: "viet copy quang cao" → kich hoat 05-copy-quang-cao.md Mot so skill goi skill khac tu dong: - 20-brief-client-intake → goi 09-insight-khach-hang + 08-nghien-cuu-doi-thu + 10-tinh-kpi-nguoc +…
The author has extensive Java/Golang/Python experience but is new to Swift and macOS app development. All generated code must include detailed comments in English explaining: - Swift-specific syntax and language features (e.g., @Observable, actor, some View, property wrappers) - SwiftUI concepts and view lifecycle (e.g., .task, .environment(), NavigationSplitView) - macOS/Apple platform APIs (e.g., NSWorkspace, DispatchSource, FileManager) - Why a particular Swift pattern is used when it differs from the Java/Go/Python equivalent First build downloads dependencies (Yams, swift-markdown, swift-collections). MVVM with…
This is the Windows Protocol Test Suites repository — a .NET-based interoperability testing framework for Microsoft Open Specification protocols. Follow these steps before starting any task: Use kebab-case for the…
This repository publishes Paper2Patent prompt templates and AI agent skills for converting academic papers into Chinese invention patent application drafts. Use .claude/skills/paper2patent/SKILL.md when the task involves paper-to-patent conversion, claim drafting, patent specification drafting, patent drawing descriptions, or fidelity checks against a source paper. Run skill validation after changes:
This directory contains every official bunqueue client and the machinery that keeps them honest. Language-specific rules live in each SDK's own CLAUDE.md; the rules here apply to ALL of them and to any new language. 1. docs/protocol.md is the wire contract. Before writing or changing ANY SDK — and especially before starting a new language — read it end to end. Every field name, response shape, clamp, and semantic rule an SDK implements MUST come from that spec (which in…
See AGENTS.md for the Zero-Trust skill loading protocol (applies to all AI agents) and the MCP runtime-enforcement section (when enabled). At the end of any multi-step task with user corrections, load and run common/common-session-retrospective to capture skill gaps and prevent repeat rework.
SPIDER (Scalable Physics-Informed DExterous Retargeting) is a physics-based retargeting framework that converts human motion (from video or mocap) into robot actions for dexterous hands and humanoid robots. Developed by Meta/FAIR.…
IMPORTANT: This is a meta-repository situation. Be aware that this repo is used to develop a system called "Claude Simone" which is an addon to Claude Code to manage Tasks and Work. At the same time this repository itself uses Simone to organize its own work (currently the mcp-server based version). You have access to the log_activity tool. Use it to record your activities after every activity that is relevant for the project. This helps track development progress and understand…
This is the MCP (Model Context Protocol) server implementation for Simone. This is a complete rewrite of the legacy directory-based command system. When configured in a project's .mcp.json:
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at http://www.apache.org/licenses/LICENSE-2.0 Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. -->
codex-drift-watch.yml runs daily on the dario-drift self-hosted runner against a dedicated live-test ChatGPT account; the DARIOCODEXLIVE_HOME repository variable holds the absolute HOME whose .dario/codex-accounts/ directory contains that account. Without the variable the workflow is inert by design. The workflow reads the account in place, masks its ChatGPT account id, and never logs or commits its tokens; test/fixtures/codex-models.snapshot.json is the seeded baseline it compares against. The maintainer reads PR bodies. Surface trade-offs and unknowns explicitly rather than picking a default and…
GodotMaker is an ECS-native text-to-game framework for Godot. It uses gecs (open-source ECS addon) with AI working in the GDScript layer. The project generates complete Godot game projects from natural…
This repository is a prompt-driven SEO audit-and-fix workflow with a runnable live audit tool. The full instructions live in the skill at .claude/skills/seo-audit/SKILL.md, which Claude Code loads automatically when a task involves search visibility — an SEO audit, indexing or crawl problems, canonicals, hreflang, sitemaps, robots.txt, structured data, a traffic drop, a migration, or AI answer-engine visibility. If the skill is not available in your environment, read START.md and follow its phases directly. Either path leads to the same workflow.
This repository serves as a curated index of agent workspace blueprints and Claude Code-related resources shared on GitHub. It provides organized, categorized listings with descriptions to help users discover relevant…
Loaded by the Claude CLI / Claude Code agent when present at the repo root. A non-test change under tinyagentos/ or desktop/src/ requires a changelog fragment in the same PR…
When building or modifying ontologies, follow this workflow. Claude decides which tools to call and in what order based on results — this is not a fixed pipeline. 6. Call ontodefects FIRST. A reasoner amplifies whatever the declarations say, so a self-contradicting ontology makes every inference suspect. Fix what it reports before materializing anything 7. Call ontodlpboundary SECOND, and read it before trusting anything ontoreason returns. onto_defects asks whether the declarations contradict each other; this asks a different question the…
A file Claude Code reads at the start of every session. It holds the commands, conventions and warnings the agent needs for this project.
Where does it go?
At the repository root. Claude Code also reads CLAUDE.md files in subdirectories when it works there.
What should it contain?
Build and test commands, the project's layout, conventions that aren't obvious from the code, and mistakes to avoid. Short files tend to work better than long ones.
CLAUDE.md or AGENTS.md?
Claude Code reads CLAUDE.md; most other agents read AGENTS.md. Many projects keep one and point the other at it.
Google DeepMindCLAUDE.md
