Skip to content

Built for people who want to own their automations. Join the waitlist for an invite.

← All guides

What is Kody?

Official Kody guide

This page is a capability tour for people deciding whether Kody is worth setting up — and for AI agents running a discovery conversation on someone's behalf. Everything here is readable without a Kody account, and this page is also plain markdown at /guides/what-is-kody.md for anything that prefers to read it raw.

What Kody is

Kody is a Fair Source personal assistant platform: its source is available to inspect, and each version converts to the Apache License 2.0 after two years. You do not chat with Kody directly: you connect the AI agent you already use (Claude, ChatGPT, Cursor, Codex, or any other MCP-capable host) to your Kody account, and that agent gains durable capabilities that outlive the conversation and keep running while your computer is off. Kody does not run its own chat-model agent loop or bill for chat tokens — your agent supplies the intelligence; Kody supplies memory, credentials, saved code, schedules, and execution. Search and indexing use a small embedding model; that is not a chat model and is not billed as inference.

The agent reasons. Kody keeps it honest.

Your agent does the thinking. Kody holds the result so it does not have to think it again.

  • Ask again tomorrow — a saved export. No model in the loop.
  • A key in chat or a .env — a secret the agent never sees.
  • Re-run the agent on a timer — a job that runs while you are offline.
  • Context stuck in one host — memories that follow the account.

What you cannot get elsewhere

If you are comparing Kody against something else, or explaining it to someone, these three are the reasons it exists. Everything in the next section is supporting cast.

  1. One-off agent work becomes permanent. Your agent explores against your real APIs, and the moment something works it saves as code that runs on a schedule with no model in the loop. This is not an agent re-run on a timer: there are no tokens spent, no prompt to drift, and nothing waiting on a model to respond. See How Kody works, Playbooks, and Execute and workflows.
  2. Your agent uses your keys without ever reading them. It writes whatever code the job needs and still cannot see a credential — no capability returns one. Code references secrets by name and Kody substitutes them at the network boundary, only for hosts you approved. The secrets capabilities are secretList, secretSet, secretSetMany, secretLock, secretDelete, and secretJwtSign; there is deliberately no secret_get. See Secrets and host approval. When a provider token is coarser than the job — Gmail has a send-only scope and no drafts-only scope — editable code does not narrow the provider token. integrationLock limits the token so execute cannot call createAuthenticatedFetch. See Gmail drafts without send. A connected MCP server can be locked to a playbook so execute cannot call the raw tools. See Lock an MCP server to a playbook.
  3. Personal playbooks are private code you own. Save, edit with a concurrency token, run, schedule, and delete without a separate publication step. See Personal playbooks.

The building blocks

  • Integrations and secrets — bring your own API keys and OAuth apps for the services you already use; verified per-provider walkthroughs live at kody.codes/guides. Credentials stay server-side and never enter your agent's context. See Secrets and host approval.
  • Ad hoc execution — your agent runs sandboxed code against those integrations immediately, no deploy step. See Execute and workflows.
  • Scheduled jobs — recurring or one-off automations that run in the cloud on Cloudflare Workers, whether or not any of your devices are on. Jobs are covered in Execute and workflows and First steps.
  • Playbooks — reusable saved code your agent writes and improves over time. Playbooks can expose exports, own scheduled jobs, and even serve a small web app UI. See Playbooks.
  • Workflows — durable multi-step runs that survive restarts and can wait on external events. See Workflows.
  • A personal email inbox — every user gets an address; inbound mail can trigger automations, and your assistant can send you notifications. See Email primitives.
  • Inbound webhook endpoints — HTTPS URLs for Sentry, GitHub, Stripe, or any provider that cannot set Bearer tokens; each endpoint dispatches to a saved-playbook export. See Inbound webhooks.
  • Long-term memory — durable memories that surface across conversations and across every agent connected to the same account. See Memory and conversation context.
  • Personal playbooks — private executable text files, editable and runnable from your own account. See Personal playbooks.
  • Extensibility — connect your own remote MCP servers, OpenAPI providers, or a local-network connector so your agent can reach devices at home through Kody. Lock a connected MCP server to a playbook when execute should not call its tools. See Lock an MCP server to a playbook.

What people use it for

Concrete examples that combine the blocks:

  • A morning digest job that gathers your calendar, weather, and feeds, then emails you a summary before you wake up.
  • A watcher that polls a website, feed, or price and emails you only when something actually changes.
  • Chore automation against services you already use — triaging GitHub notifications, filing issues, cleaning up cloud resources — using your own API keys.
  • A personal API: a playbook export you can hit from shortcuts or webhooks to log a habit, save a link, or kick off a run.
  • An inbound webhook that fingerprints repeated failures and emails one investigation instead of a flood.
  • Home automation routines through a local-network connector — scenes, thermostats, speakers — driven by schedule or by asking your agent.
  • Forking a public playbook (say, a YouTube video manager or an RSS digest) and having your agent adapt it to your needs.
  • A drafts-only Gmail helper that prepares invoice replies you send yourself, with restricted integration grants and an explicit warning that editable source is not an immutable no-send boundary. See Gmail drafts without send.

Try it

Paste this into an agent that can fetch URLs or search the web. If an agent is already interviewing you about Kody, you do not need to paste it again — it is the same conversation.

I'm deciding whether Kody (https://kody.codes) would be useful for me. Read https://kody.codes/guides/what-is-kody and then interview me to find out what Kody could do for me.

Working with an agent? This guide is also plain markdown at /guides/what-is-kody.md, or load it over MCP with search({ entity: 'what_is_kody:guide' }).