Teakit

LLM-friendly TypeScript Kits. Small, typed, and composable utilities designed for developers and AI coding agents.

kit.ts

typed kit

import { defineKit } from "@teakit/core"; export const reportingKit = defineKit({  name: "asset-reporting",  goal: "turn source data into reviewable notes",  input: {    dataset: "portfolio-q2",    question: "NOI variance",  },  output: "cited summary with next actions",});

Small TypeScript kits for serious AI-assisted work

Teakit turns recurring product and engineering patterns into typed, readable kits that humans can maintain and LLMs can use with less drift.

Readable by people and agents

Small typed modules keep purpose, inputs, outputs, examples, and review rules visible.

Context 2

Context stays close

Keep prompts, schemas, examples, constraints, and tool contracts next to the code that uses them.

Reuse 3

Reusable by default

Turn repeated workflows into kits that can be copied, tested, adapted, and improved over time.

Context

From scattered intent to reusable TypeScript kit

Put goals, inputs, examples, constraints, and expected outputs in one typed module so agent work has a stable surface to follow.

Read Teakit docs

workflow.ts

import { runKit } from "@teakit/core";import { reportingKit } from "./kits/reporting"; await runKit(reportingKit, {  dataset: "portfolio-q2",  question: "Explain NOI variance",  audience: "asset review meeting",});

Kit parts

Keep the useful parts close

Teakit docs

prompts

schemas

tools

examples

reviews

agents

docs

tests

memory

workflows

adapters

contracts

01

Name the job

Define what the kit is for, what it needs, and what result should be easy to review.

02

Attach context

Keep schema, examples, constraints, and operating notes in TypeScript where agents can inspect them.

03

Reuse the kit

Run the same kit across product code, docs, internal tools, and agent workflows.

LLM-friendly TypeScript Kits.

Teakit turns recurring product, agent, and engineering patterns into small, typed, composable utilities that developers can maintain and AI coding agents can reuse.