token-efficiency

Claude Code skills for developers who code like cats — never more effort than the problem requires.

Your project remembers across coding agents. Compact, cited project memory for Codex, Claude Code, Cursor, and compatible tools.

Unofficial lightweight Superpowers skills overlay for GPT-5.6, Claude Fable 5, and modern coding agents

A lightweight file-mapping standard that lets any AI tool navigate large codebases without reading every file. One index, two rules, any AI. No more wasted tokens, no more wasted time.

Honey ESON (Efficient Structured Object Notation) — compact, lossless wire encoding for agent-to-agent structured payloads. Spec, JS + Python reference implementations, conformance vectors, LLM primer, benchmarks. MIT.

Keep your development alive and never lose context.

Convert MCP servers into Claude Code skills — not MCP connections. Token-minimal, code-execution pattern.

Codex plugin MCP server (codex-plugin-mcp) for token efficiency, token reduction, context compression: exact-recoverable terminal, file, project, web, memory.

Cut your GitHub Copilot token bill, locally. Certance Field compresses oversized context before it reaches the model, redacts and scans for secrets, and exports audit evidence. No cloud, no external calls.

Claude Code skill: caveman-style token efficiency, spoken with a real Spanish regional accent

网页系统侦察 · DeepSeek Harness 插件：摸清一个网页系统怎么运作，只摸一次。抓真实接口与可访问性树，固化成可复用的作战手册。零依赖，不用 Playwright。

Route each prompt to the smallest model strong enough — heuristic+judge tier routing, multi-provider, response cache, budget caps.

A Codex plugin for professional brevity, selective model routing, and bounded agent collaboration.

Clad - a token-minimal programming language designed for LLM code generation

🪨 Modo ultra-comprimido PT-BR para Claude Code — fork profissional do caveman com acentos obrigatórios, emojis preservados e safety carve-out expandido. Corta ~75% tokens.

The neural insulation layer for AI coding agents. 93% token reduction, 6× longer sessions. Serena + Semble + Headroom + RTK + mem0 in one installer.

til (тіл) — a scripting language engineered for AI agents: the whole spec is a 1.8k-token prompt card, token parity with Python, errors built for self-repair. Playable Flappy Bird included.

Deterministic dual-edge code-context graph for AI coding sessions — code edges + capability edges (string-keyed registries imports can't see). Zero LLM tokens. AGPL-3.0.

Zero-dependency MCP server that cuts round-trips and tokens for coding agents: 6 aggregate tools (agent_search, agent_read, agent_edit, agent_exec, agent_git, agent_savings)

🔎 Token-efficient code & doc retrieval for agents: FTS5 search, symbol outlines, surgical reads. CLI + web + MCP. Zero deps.