Developing in worktrees

You can run several copies of one factory on one machine, each with its own environment.

A trigger or schedule that reads active from the environment, as the examples do, stays off in a new copy until that copy turns it on.

Once per engineer: a dev factory on the hub

1. In the hub, under Factories, add a factory named for you, such as alice-dev.
2. On its Settings tab, connect the same apps as your production factory.
3. Keep the token the hub shows; one running copy uses it at a time. Two copies with one token split its events between them.

A copy with no active trigger, whose workflows and bindings use none of these providers, needs no token.

For each worktree

Say production answers questions in a Slack channel, and you are changing that. The trigger reads both its flag and its channel from the environment:

import { slack } from "@jigs-ai/jigs";

// In defineFactory's `triggers`.
const triggers = {
  answer: {
    active: process.env.ANSWER_ACTIVE === "true",
    workflow: "answer",
    source: slack.mentions({
      installationName: "slack-acme",
      channels: [process.env.ANSWER_CHANNEL ?? ""],
    }),
  },
};

Production's channel can live in the shared .env; a copy's .env.local overrides it.

From the main checkout:

git worktree add ../my-factory-answer
cd ../my-factory-answer
pnpm install
cp ../my-factory/.env . && cp .env.local.example .env.local

.env holds the values every copy shares, so copy it as it is. In .env.local, change COMPOSE_PROJECT_NAME and the three ports, and the port in WORKFLOW_POSTGRES_URL, to ones no other copy uses. Then add your dev factory's token and turn on the one trigger you are working on, in a test channel:

# .env.local
COMPOSE_PROJECT_NAME=my-factory-answer
JIGS_SERVICE_PORT=8899
JIGS_DASHBOARD_PORT=9199
JIGS_POSTGRES_PORT=5499
WORKFLOW_POSTGRES_URL=postgres://jigs:jigs@localhost:5499/jigs
JIGS_HUB_TOKEN=<your dev factory's token>
ANSWER_ACTIVE=true
ANSWER_CHANNEL=C0TESTCHAN

Then start it:

pnpm exec jigs up

jigs status lists every trigger and schedule with its STATE, so you can check that only answer is active.

Stop the copy with pnpm exec jigs down before you remove the worktree. Its Postgres data stays in the Docker volume <COMPOSE_PROJECT_NAME>_postgres-data until you remove it with docker volume rm.