Tiered model routing for token economy: the strongest model thinks, cheaper models grind. Ready-made subagents (sonnet scout, implementer, task-reviewer and E2E runner, haiku surveyor, test-runner and verifier, opus reviewer) plus a routing skill and dispatch stats.
★not rated 4
changed 29d agoA
tokens not measured
originalMIT
Turn Claude into a C++26 specialist — reflection, contracts, senders, #embed, expansion statements. Standard-first: suggests the C++26 form even when clang/gcc lag.
★not rated 4 5mo agoA
tokens not measured
originalMIT
Recursive Language Model integration for Claude Code using rlm-rs CLI. Enables processing documents far exceeding context windows via chunking and recursive sub-LLM calls.
Commands and skills that hold a coding agent to one owner's standards. /sonu:build designs a change with you, builds it test-first and hands it back with the riskiest parts listed. /sonu:ship takes a finished change to a merged PR: it gathers the repo's AI reviewer bots, fixes or answers each finding, and merges when…
Fetch relevant, high-quality, free-to-use images from the web for any project context. Scans your directory for context or takes a description, searches Unsplash/Pexels/Pixabay (API or zero-config WebSearch fallback), presents curated picks, and downloads to your project.
★not rated 4 8mo agoA
tokens not measured
originalMIT
Universal AI skill router with dual-mode discipline: ponytail for code, caveman for chat. Auto-routes any prompt to the best installed skill. Zero slash commands. 75% fewer tokens. Works with Claude Code, Codex, Cursor, Gemini CLI, OpenCode.
SessionStart and SubagentStart hooks that inject the GraphSmith multi-agent discipline (from the repo's canonical SKILL.md) into context for every session and every Task-spawned subagent from the moment it starts.
★not rated 4 1mo agoA
tokens not measured
originalMIT
Turns Claude Code into a full product-development toolkit: an opinionated plan → verify → commit engineering workflow, specialized subagents for legal compliance, UX design, backend/frontend/cloud architecture, Go/Java/TypeScript implementation, DevOps/infrastructure (Terraform, Kubernetes, Docker, AWS/Azure/GCP), and.
★not rated 4
changed 1mo agoA
tokens not measured
originalMIT
CRISP design intelligence for AI agents: design reviews and audits with A-F grades, accessibility, AI surfaces, motion craft, funnels, DESIGN.md study, before/after proof, and a design detector that checks every UI edit.
JDB (Java Debugger) for Claude Code via MCP - attach to running JVMs and debug with breakpoints, expression evaluation, thread inspection, and class introspection.
★not rated 4 1mo agoA
tokens not measured
originalApache-2.0
Interactive architectural review through questioning - never assumes, always asks. MUST BE USED when user mentions 'architecture', 'design patterns', or implementing features. Use PROACTIVELY for codebase review.
Claude Code plugin: research workflow — source logging, source summarization, and deep-dive primitives, with deep-research/technical/osint/georeaction/stack/ecosystem/competitor/purchasing/general-research-workspace/obsidian-vault variants. The obsidian-vault variant scaffolds the same research loop as a working…
WORK IN PROGRESS — KDE Plasma (Wayland) runtime utilities. The desktop-environment half of a distro × DE pairing: KWin scripting, plasmoid management, panel layout, virtual desktops & activities, KGlobalAccel shortcuts, theme/look-and-feel switching, KDE Connect, Klipper, Baloo, kwriteconfig, qdbus introspection.…
★not rated 4
changed 6d agoA
tokens not measured
originalMIT
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: