sonnet-opus-advisor-haiku-subagents
A Claude Code settings rule that sets Sonnet as the main model, Opus as an advisor consulted at decision points, and Haiku for subagents.
Resources that help an AI assistant read, write, review and refactor code.
41,654 results
A Claude Code settings rule that sets Sonnet as the main model, Opus as an advisor consulted at decision points, and Haiku for subagents.
A rule that sets Claude Sonnet 5.5 as the main model for day-to-day coding, pinned to its full model ID.
A partnership settings rule by davila7 that connects the GLM-4.6 model from the GLM Coding Plan with Claude Code and more than ten other AI coding tools.
A permissions rule file that pre-approves common everyday git commands for version control work in Claude Code.
A Claude Code permissions rule that asks for approval before Stripe MCP tools write data, move money or start a Stripe Atlas company formation.
A statusline configuration by davila7 that the publisher describes as tracking asset processing status for game development workflows.
A statusline configuration by davila7 that shows a dice-based chip counter during a coding session.
A space-themed status line that tracks a ship's fuel, warp level, distance and condition as coding activity accumulates.
A statusline configuration that shows build status for a game across iOS, Android, PC and WebGL targets.
A monitor that shows Vercel build status and time since the last deployment, with a clickable link to the deployment URL.
A rule-type configuration that monitors Vercel deployments and builds for failures and sends desktop notifications when errors are detected.
A minimal Claude Code status line showing project name, git branch, model and context usage, with worktree-aware naming.
A skill by davila7 that covers production deployment principles, decision-making, safe deployment workflows, rollback strategies and verification.
A rule that sets LangSmith tracing environment variables so Claude Code conversation traces can be sent to LangSmith for monitoring and analysis.
A Claude Code skill that queries a code graph database to show component relationships, dependencies and the likely impact of a change.
A skill covering LlamaIndex, a data framework for building LLM applications with retrieval-augmented generation (RAG), document ingestion, indexing and querying.
A skill that compiles a daily digest of new AI releases from lab repositories, a changelog, a blog and Hacker News, and sends it by email.
A skill for building AI agents that view screens, move the cursor, click buttons and type text, with guidance on Anthropic Computer Use, OpenAI Operator/CUA and open-source alternatives.
A skill covering PyTorch Lightning, a high-level PyTorch framework that handles distributed training with a Trainer class and a callbacks system.
A skill for PyTorch-native distributed LLM pretraining with torchtitan, using 4D parallelism across FSDP2, TP, PP and CP.
A skill covering LLaVA, a vision-language assistant that pairs a CLIP vision encoder with Vicuna or LLaMA language models for image-based conversation.
A skill by davila7 for LLM observability, covering tracing, evaluation and monitoring of AI applications, released under the MIT license.
A skill by davila7 for activation-aware 4-bit weight quantization of large language models, published under the MIT license.
A skill by davila7 for designing database schemas for SQL and NoSQL databases, covering normalization, indexing, migrations, constraints and performance.