MeshHook is an MIT-licensed, webhook-first workflow engine with a visual builder (SvelteKit/Svelte 5) and Temporal-like durability via event sourcing on Postgres.
- UI/API: SvelteKit (Svelte 5)
- Database: Postgres, through
@profullstack/libsql-pg(the queries kept their libSQL shape) - Workers: Node.js (or Bun), stateless
- Queue: SQLite-backed, with visibility timeouts and a dead-letter queue
- Auth: self-hosted sessions (scrypt password hashing, opaque session tokens)
- Live logs: Server-Sent Events
- Transforms: JMESPath
| Feature | n8n | Windmill | Temporal | MeshHook |
|---|---|---|---|---|
| License | Fair-code (restrictive) | AGPLv3 | MIT | MIT |
| Primary Use Case | No-code automation | Script orchestration | Microservice workflows | Webhook-first workflows |
| Durability | Database polling | Database + queues | Custom event sourcing | Event sourcing on SQLite |
| Visual Builder | ✅ Drag-and-drop | ❌ Code-first | ❌ Code-first | ✅ Visual + Code |
| Webhook-Native | ❌ Not primary | ✅ Built-in | ||
| Self-Hosted | ✅ Yes | ✅ Yes | ✅ Yes | ✅ Yes |
| Database | MySQL/Postgres | Postgres | Custom | Postgres |
| Transforms | JavaScript | TypeScript/Python | Any language | JMESPath |
| Realtime Logs | ❌ Polling | ❌ Polling | ❌ Polling | ✅ Server-Sent Events |
| Deployment | Complex (multiple services) | Complex (workers + API) | Complex (server + workers) | Simple (single service) |
| Learning Curve | Low (no-code) | Medium (scripts) | High (SDK required) | Low (visual + simple) |
- Webhook-first: Built specifically for webhook orchestration and transformation
- Simple deployment: Single service on one port, no complex infrastructure
- MIT licensed: Truly open source, use anywhere without restrictions
- Event sourcing: Temporal-like durability without the complexity
- Visual + Code: Low-code visual builder with JMESPath for power users
- No infrastructure: local development is a single SQLite file — no server, no containers
-
Install dependencies
bun install
-
Setup environment (interactive)
bun run setup
Select "Local Development" when prompted. This creates
.env.localpointing at a local SQLite file (file:./meshhook.db) and generates a secrets encryption key.There is no database server to start — the file is created by the migration step.
-
Run migrations
bun run db:migrate
-
Start the orchestrator
bun run start
Runs on port 8080 (configurable via PORT environment variable)
-
Install dependencies
bun install
-
Setup environment (interactive)
bun run setup
Select "Production" or "Staging" and enter your Postgres URL (
postgres://user:pass@host:5432/meshhook). Anything else is refused. -
Run migrations
bun run db:migrate
Applies any pending migrations from
migrations/and records them inschema_migrations. Re-running is a no-op. -
Start the server
bun run start
Runs on port 8080 by default. Configure your reverse proxy to forward traffic to this port.
bun run setup- Interactive environment configuration (local/staging/production)bun run db:migrate- Apply pending database migrationsbun run db:status- Show applied and pending migrationsbun run db:verify- Verify the schema matches what the app expectsbun run test- Run the test suitebun run start- Start the orchestrator worker (production)bun run mh --help- CLI help
.env.local- Local development (committed to repo with safe defaults).env.staging- Staging environment (not committed).env.production- Production environment (not committed).env- Symlink to active environment (created by setup script)
./docs/Environment-Setup.md- Detailed environment setup guide./docs/Turso-Migration.md- What changed in the move off Supabase (to Turso, since superseded by Postgres: seemigrations-pg/)./docs/PRD.md- Product requirements./docs/Architecture.md- System architecture./docs/diagrams/*.puml- PlantUML diagrams
MeshHook runs as a single service on port 8080 that handles:
- Webhook intake and processing
- Workflow orchestration
- Background job execution
- HTTP request execution with retries
All components communicate through the Postgres database - no inter-service HTTP calls needed.