@hilmo/bruno-openapi-sync

bruno-openapi-sync

Sync an OpenAPI spec into a Bruno collection (OpenCollection YAML) while keeping manual changes like variables or scripts.

Fork of brpaz/bruno-openapi-sync, published to npm as @hilmo/bruno-openapi-sync.

Motivation

Bruno collections built from an OpenAPI spec tend to get hand-edited over time — auth, test scripts, example bodies. Existing generators regenerate the whole collection on every run, so a re-sync after the spec changes wipes that work out. This project exists to make re-syncing safe: keep structure in lockstep with the spec while leaving anything a human touched untouched.

AI-assisted / vibe-coded project, built for personal use. This tool was built end-to-end with an AI coding agent, from requirements-gathering through implementation and tests, with a human reviewing decisions and steering scope along the way. It's tested (30 tests, all passing) and has been run against a real hosted spec, but it hasn't seen production use. Use it at your own risk — read the code before trusting it with a collection you care about.

Why not Bruno's own built-in OpenAPI sync?

Bruno's Electron app has a native "sync collection from OpenAPI source" feature (bruno.json's openapi config: sourceUrl, groupBy, autoCheck, etc.) — but it's gated behind a paid plan. This tool is a free, open, CLI-based alternative: no app-side license check, works in CI/scripts, and (being closer to the metal) can implement the field-level ownership model above rather than a simpler regenerate-on-check.

Quick Start

npx @hilmo/bruno-openapi-sync sync --spec https://petstore3.swagger.io/api/v3/openapi.json --output ./my-collection

Open ./my-collection in Bruno. Change the spec, run the same command again — your edits stay, the structure stays in sync.

Features

• Safe re-sync — generated fields (method, url, folder placement) always update; user-owned fields (runtime scripts, hand-added headers, edited auth/body) never do
• Field-level, not file-level, ownership — a single request file can mix spec-derived and hand-written content; sync only ever touches the parts it owns
• Orphan handling — a request whose operation was removed from the spec is deleted automatically if untouched, or kept with a warning if you've added anything to it
• Folder sync — requests move between folders automatically when an operation's tag changes
• --overwrite escape hatch — when you do want a clean, forced regeneration, one flag suspends every ownership rule
• --dry-run preview — see exactly what would change (create/update/move/delete/skip) before anything is written
• No hidden state in your request files — match-key tracking lives in a sidecar (.bruno-openapi-sync/state.yml), never in a field Bruno's own app might strip

Installation

Pre-requisites

• Node.js ≥ 22

From npm

npm install --global @hilmo/bruno-openapi-sync
bruno-openapi-sync sync --spec ./openapi.yaml --output ./my-collection

Or skip the install and run it directly with npx:

npx @hilmo/bruno-openapi-sync sync --spec ./openapi.yaml --output ./my-collection

From source

Additionally requires pnpm ≥ 11.

git clone <this-repo>
cd bruno-openapi-sync
pnpm install
pnpm build

dist/cli.mjs is then a runnable Node ESM script. Link it locally to use the bruno-openapi-sync command anywhere:

pnpm link --global
bruno-openapi-sync sync --spec ./openapi.yaml --output ./my-collection

Usage

bruno-openapi-sync sync --spec <path-or-url> --output <dir> [--overwrite] [--dry-run]

Flag	Required	Description
--spec <pathOrUrl>	yes	Local file path or http(s):// URL to an OpenAPI 3.0.x/3.1.x spec (Swagger 2.0 is rejected)
--output <dir>	yes	Bruno collection directory. Created/bootstrapped automatically if it doesn't exist yet
--overwrite	no	Suspend all field-ownership rules — the spec wins everywhere, including fields you've hand-edited
--dry-run	no	Print the plan (creates/updates/moves/deletes/skips) without writing anything

Examples

# First sync — creates the collection from scratch
bruno-openapi-sync sync --spec ./openapi.yaml --output ./my-collection

# Re-sync after the spec changed — updates structure, keeps your edits
bruno-openapi-sync sync --spec ./openapi.yaml --output ./my-collection

# Preview what a sync would doapidevtools/swagger-parser
bruno-openapi-sync sync --spec ./openapi.yaml --output ./my-collection --dry-run

# Force a clean regeneration, discarding hand-editsapidevtools/swagger-parser
bruno-openapi-sync sync --spec ./openapi.yaml --output ./my-collection --overwrite

Sample output:apidevtools/swagger-parser

Did update pets/List all pets.yml
Did move pets/Find pet by ID.yml -> animals/Find pet by ID.yml
Did delete pets/Create a pet.yml
Did skip deleting Health check.yml (has a runtime block (scripts/tests/assertions/variables))
1 created, 2 updated, 1 moved, 1 deleted, 1 skipped.
warning: Kept orphaned request at Health check.yml (operation "getHealth" is no longer in the spec) — has a runtime block (scripts/tests/assertions/variables).

How field ownership works

Every request/environment file is split into three kinds of field:

• Generated — info.name, http.method, http.url, folder placement, and per-entry-matched http.params/http.headers. Always overwritten from the spec on every sync.
• Seeded — http.auth, http.body, info.seq, and environment files. Given a best-effort default the first time a request is created, then left alone on every later sync — even if you edit them, even if the spec changes.
• User-owned — everything else: runtime scripts, settings, custom headers/params you added by hand, any extra key you add to the file. Never touched.

The match key linking a spec operation to its file across syncs is the spec's operationId, tracked in .bruno-openapi-sync/state.yml — never written into the request file itself (Bruno's own schema validation can't be trusted to preserve an unrecognized field through an open-and-save round trip).

Full rationale is in CONTEXT.md and docs/adr/.

Development

pnpm install
pnpm test        # vitest
pnpm typecheck    # tsc --noEmit
pnpm build        # tsdown -> dist/

Tests live in test/, fixture OpenAPI specs in test/fixtures/specs/. The whole tool is exercised through one library seam — sync() in src/index.ts — which the CLI and every test call directly.

Project status

Feature-complete against its original spec (.scratch/openapi-sync/spec.md): spec loading (local/URL, 3.0/3.1 only), first-sync collection creation, safe re-sync, per-entry param/header diffing, orphan handling, folder-move-on-retag, and --overwrite/--dry-run. See .scratch/openapi-sync/issues/ for the ticket-by-ticket history.

License

MIT — see LICENSE.