Plugins

Claude Code の hook 経由で圧縮プラグイン (rtk/headroom) の効きを観測・記録する計測機構。caveman は prompt mode で観測不可だが /token-report で transcript 推定値として併記

Structured procedures for optimizing Anthropic model cost and usage consumption in every session — model routing, context hygiene, caching, batching, and usage audits.

Optimisation tokens — routing Haiku/Sonnet/Opus, classification T1/T2/T3, compression caveman, budget tracking. Inclut 6 skills + 3 agents.

Per-operation token usage reporter for Claude Code. Shows token counts, cost estimates, tool attribution, cache efficiency, duration, bash commands, web fetches, and file activity when agents complete. Only outputs in debug mode (claude --debug).

Acumula tokens e custos por projeto, com histórico diário e gráficos (ASCII e HTML)

Tokenmin — Apache-2.0 scanner + engine that audits how you use Claude and recommends concrete improvements. Reads ~/.claude, anonymizes locally, never sends anything off-machine by default.

Corpus-aware context packing, safety hooks, and productivity skills for Claude Code. Works with or without the tokenpak proxy.

Deterministic transcript-JSONL cost analyzer — per-message cost curve, 5h rolling window, floor breakdown, and cache-read vs cache-write vs fresh-input split. Prerequisite for Part C (status-line meter) and Part G (heat-based floor).

Active performance and cost engineering for AI development loops.

连接本机同花顺，在 Claude Code 中查询行情、持仓并使用实时盯盘。

Engineering + Product team — 23 agents, 2 teams. All agents and skills in one install.

23-agent engineering + product team with 125 skills for Claude Code

LLM-powered knowledge graph for academic papers: tracks concepts, projects, and engagement as you read, with a local Too Many Papers web UI.

Bundles two agent skills: challenger (adversarial stress-tester for technical decisions, designs, and specs) and skill-forge (turns repeated manual workflows into reusable, validated Agent Skills).

Scaffold a new TypeScript CLI tool's documentation and configuration, or audit an existing tool against the project tool-conventions (docs/tools/<name>.md format, ~/.tool-agents/<name>/ config folder, four-tier env-var resolution chain, vendor-canonical LLM provider names, no-fallback rule). Bundles the /tool-conventions command and its tool-doc-config-architect subagent.

Automatisierte Skill- und Plugin-Verbesserung: Strukturpruefung, CLAUDE.md-Audit, A/B-Benchmark mit Varianz-Analyse und HTML-Report

Evaluate and compare developer tools with structured scoring criteria

基于动态 Gate、分级证据、场景权重和敏感性分析调查、比较并筛选最适合用户约束的工具

Find, compare, improve, self-check, and plan installation for agent tools across callable Codex tools, connector plugins, local inventory, live GitHub, Hugging Face, skills.sh, npm, Python, Homebrew, and deferred package registries.

找出任務可用的工具（MCP、Skills、CLI）或搜尋更好的替代方案

Intelligent permission gate for shell, file, search, Skill, and MCP tool surfaces. AST-parses Bash commands, guards file reads/writes, and blocks configured tool invocations. Tracks manually approved commands for permanent rules.

Register the deferred tool_search MCP bridge

from eliaslecomte/claude-plugin

Personal utility commands and tools for Claude Code

Infrastructure and authoring utilities for agent workspaces, handoffs, audits, harvesting, and plugin management

Docker, Drizzle ORM, pnpm workspaces, Zod schemas, Claude Code plugin conventions, npx skills conventions

Sean O'Dell's personal Claude Code toolkit — GitHub SSH key management and more

Everyday toolkit: git workflow skills, TDD hooks, code-review helpers, Forgejo CLI, SignalDB observability, dashboard design, and GitHub issue/PR commands.

Plugin from SergioLNeves/toolkit-agent-skills

A team of five specialist agents coordinated by one skill. Activates automatically on messages wrapped in [[ double square brackets ]]. Handles prompt engineering, guardrails, and testing/evaluation automatically; asks before taking on coding or skill-building tasks.