emt

Prompt & context engineering for agents as a Claude Code plugin — sharpen requests, forge agent system prompts, review through expert lenses. Zero deps, no API keys.

Zero-dependency, fully auditable TypeScript library that scores a RAG answer inline at query time. No LLM calls. No offline pipeline. Uses signals already in your retrieval stack. Output: a typed 0–100 scorecard with per-dimension breakdowns.

Recursive-fold consolidation for append-only agent memory — evolving fixed-size digests so memory gets denser as it grows, not just longer. Built to interoperate with claude-mem.

A time tracking and billing plugin for the claude-for-legal ecosystem. Captures billable attorney time per client and matter in 6-minute increments, supports multi-attorney firms via shared folder, generates printable invoice exhibits, and includes WIP review, budget tracking, and a monthly billing digest agent.

A tiny remote MCP server that carries a context pack (voice rules, system facts, anything) between your Claude surfaces: push from Claude Code, read from Claude mobile.

100+ AI Agent & RAG apps you can actually run — clone, customize, ship.

`ledger` is a local-first, queryable record of architectural decisions across projects. It is a Python CLI tool backed by SQLite, with optional AI-powered retrieval (Layer 2) and an MCP server surface

Autonomous AI operations agent for small law firms. Daily Closeout Brief, deadline escalation, billing WIP review, pre-bill scrubber, client comms drafts, anomaly detection. Built on Google ADK + Gemini 2.5 Pro + Firestore + Cloud Run.

🪨 why use many token when few token do trick — Claude Code skill that cuts 65% of tokens by talking like caveman

A suite of plugins for legal workflows

The agent that grows with you