[!IMPORTANT]
Trenova is pre-release software under active development. It is not yet suitable for production use. Follow releases or join the waitlist for updates.
What is Trenova?
Trenova is a transportation management system (TMS) for small and mid-sized over-the-road carriers in the United States. Enterprise TMS platforms often cost more than $50,000 a year, which puts them out of reach for the carriers that make up most of the industry. Trenova aims to close that gap with a flat-priced, self-hostable TMS that covers the full order-to-cash cycle: shipment entry, dispatch, driver assignment, rating, billing, settlement, accounting, and compliance.
The source is public under the Fair Source Functional Source License, so carriers and integrators can read the code, audit it, and run it on their own infrastructure.
Features
- Dispatch and planning: shipment lifecycle, stop sequencing, driver and equipment assignment, exception management.
- Rating and accessorials: formula-based rate calculation, fuel surcharge, detention and accessorial rules.
- Billing and invoicing: billing readiness checks, invoice generation, adjustments, and cash application.
- Accounting: general ledger, close workflows, and settlement / driver pay.
- Document automation: BOL, POD, rate confirmation, invoice, and weight-ticket extraction and templated document generation.
- Compliance: FMCSA-oriented compliance records, driver certification tracking, and audit trails.
- EDI and integrations: EDI 204 / 210 / 214 / 990 / 997 partner workflows, telematics (Samsara), PC*MILER, AS2.
- Search and real-time: Meilisearch-backed search and Redis-backed real-time updates fed by change data capture.
- Identity and access: OIDC sign-in, role-based permissions, and per-organization tenancy.
See the full capability list at trenova.app/features.
Tech stack
| Layer | Technology |
|---|
| API | Go 1.27, Gin, gqlgen GraphQL |
| Database | PostgreSQL via pgx, SQL-first migrations |
| Search and cache | Meilisearch, Redis (JSON and Streams) |
| Change data capture | GTC: a Go PostgreSQL logical-replication connector that projects rows into Meilisearch and Redis (runs as the ghcr.io/emoss08/gtc image; routing in deploy/gtc/sinks.yaml) |
| Web client | React 19, TypeScript, Vite, GraphQL |
| Documents | Gotenberg (PDF), docling-serve (document extraction), MinIO / S3-compatible object storage |
| Deployment | Docker images on GHCR, Caddy, Docker Compose |
Repository layout
services/tms Core TMS: API, worker and CLI (Go)
client/apps/web Web application (React)
client/apps/dash Driver portal (React)
client/packages/ Shared TypeScript packages: design system, GraphQL client, config
desktop/capture Windows companion for scanning and printing (Rust)
shared/ Shared Go packages (money, geo, dispatch planner, PC*MILER, AS2, ...)
i18n/ Translation catalogs and the extraction tooling
dev/ Local environment: compose.yml, its configs, Samsara and EDI partner simulators
deploy/ Dockerfiles, self-host compose, Caddy, GTC routing, observability, PgBouncer example
tools/ Tooling that is not shipped: extraction fine-tuning (Python), screenshots, load tests, release
docs/ Engineering guides (indexed in docs/engineering/README.md), product guide, self-hosting
Quick start (local development)
Prerequisites: Go 1.27.1, Node.js 26 (client/.nvmrc) with pnpm, Docker, and Task.
git clone https://github.com/emoss08/trenova.git
cd trenova
cp .env.example .env
# Start PostgreSQL, Redis, MinIO, Meilisearch, Gotenberg, and docling-serve
docker compose -f dev/compose.yml up -d
# Run the TMS API
task tms:run
# Run the web client
cd client && pnpm install && pnpm dev
Low-resource machines
If a build uses all your RAM and CPU, turn on low-resource mode. Go then compiles one package at
a time on one thread and runs its garbage collector harder, golangci-lint and the test runners
are capped the same way, hot reload waits 2 s after the last save, and docker-up starts only
PostgreSQL, Redis, MinIO and Temporal. Builds take longer, but the machine stays usable.
export LOW_RESOURCE=true # every task from now on (add it to your shell profile)
task tms:run-watch LOW_RESOURCE=true # or a single command
task tms:docker-up-minimal # just the minimal containers, in any mode
On Windows, setx LOW_RESOURCE true sets it for every new terminal.
LOW_RESOURCE_JOBS=2 allows two packages at a time with two threads each; that is a sensible
setting on a 16 GB machine. Run only the client app you are working on
(pnpm --filter @trenova/web dev) rather than pnpm dev, which starts both.
Malware scanning locally
ClamAV is in the local stack behind a profile, since it reserves 3 GB of memory and may use
4 GB while it reloads its signatures. Start it, wait for it to report healthy (the first start downloads its signatures, which
takes several minutes), and point the TMS at it:
docker compose -f dev/compose.yml --profile clamav up -d clamav
docker compose -f dev/compose.yml ps clamav # wait for "healthy"
# services/tms/config/config.yaml
contentScan:
clamav:
address: localhost:3310
This is the beginning of the README. Read the rest at the original. Read the full README ↗
