The README prerequisites were both stale, and the docs restated the Go
version in three places.
## Prerequisites: name the source, do not restate the version
| README said | Reality |
|---|---|
| `Go 1.24.x` | `go 1.26` in
[go.mod](https://github.com/modelcontextprotocol/registry/blob/main/go.mod)
|
| `golangci-lint v2.4.0` | `v2.11.4` pinned in
[ci.yml:30](https://github.com/modelcontextprotocol/registry/blob/main/.github/workflows/ci.yml#L30)
|
Rather than swap in fresher numbers — which only resets the clock on the
same drift — this **removes the numbers** and names the authoritative
file:
```
- **Go** - version per the `go` directive in go.mod; the Go toolchain fetches it for you
- **golangci-lint** - version per the `golangci-lint-action` pin in ci.yml, so local lint matches CI
```
The Go version was stated in **three** places, which is exactly why all
three drifted together. All three now point at a single source, so the
prerequisites list cannot go stale.
Two notes on why the numbers earn so little in prose:
- **Go corrects itself.** `go.mod` has no `toolchain` directive, so
under the default `GOTOOLCHAIN=auto` the toolchain downloads the
required version automatically. A contributor on an older Go gets an
upgrade, not a failure — it only hard-errors under `GOTOOLCHAIN=local`.
(An earlier version of this description said such a contributor "cannot
build"; that was too strong.)
- **golangci-lint does not.** Version skew here genuinely changes what
you see: running CI's config under a newer golangci-lint (2.12.2)
surfaces 23 `goconst` findings in
`cmd/publisher/commands/validate_test.go` that CI's pinned 2.11.4 does
not report. That is why the README now points at the pin rather than a
number someone has to keep in sync.
## `make check` does more than documented
Documented in three files as "lint, unit tests and integration tests".
Actually [`check: dev-down lint validate
test-all`](https://github.com/modelcontextprotocol/registry/blob/main/Makefile#L85)
— it also runs schema validation and **stops your running `dev-compose`
environment**. Worth knowing before running it mid-session. The
Makefile's own `##` help string already said this correctly; only the
prose was stale.
## Other verified corrections
- **README project structure** omitted `cmd/registry/` — the server
entry point and ko/goreleaser build target — and `internal/importer/`.
- **`tests/integration/README.md`** said the suite launches **MongoDB**.
It launches PostgreSQL (`postgres-integration-test`), as the same file
admits nine lines later. It also cited `/v0/servers/{server_id}`, a path
removed a while ago; the test actually calls `GET /v0/servers` and `GET
/v0/servers/{serverName}/versions/latest`
([main.go:210,255](https://github.com/modelcontextprotocol/registry/blob/main/tests/integration/main.go#L210)).
- **`scripts/mirror_data/README.md`** pointed at
`internal/database/migrations/008_separate_official_metadata.sql`, which
does not exist — that migration is `009_`, while `008_` is
`clean_invalid_data.sql`. (Migration 008's own header still calls itself
"007a", which is how the numbering slipped.) Also swapped two "edit line
NN" pointers for identifier names, since both had already drifted
(`maxMigration` is at :30 not :24; the connection string at :23 not
:17), and corrected the fetch URL from `/v0.1/servers` to the
`/v0/servers` the script actually requests.
- **`.env.example`** was missing
`MCP_REGISTRY_ENABLE_REGISTRY_VALIDATION`, even though the config reads
it
([config.go:18](https://github.com/modelcontextprotocol/registry/blob/main/internal/config/config.go#L18))
and README:50 tells contributors to set it while pointing at
`.env.example` as *the* reference.
- **`docker-compose.yml`** set `MCP_REGISTRY_ENVIRONMENT`, which nothing
reads — no `ENVIRONMENT` env tag in `internal/config`, and no other
occurrence in the repo. Removed. (The Pulumi-side
`mcp-registry:environment` key is a separate, real thing and is
untouched.)
- **Deleted `CHANGES.md`** — 0 bytes, untouched since the squashed
history base, zero references anywhere.
## Verification
`docker compose config` validates. `grep` for `Go 1.x` / `golangci-lint
vX` across all `.md`/`.mdx` now returns nothing outside `complete.md`
(which #1522 deletes). Both `../../README.md#pre-requisites` links
resolve and the anchor matches the real `#### Pre-requisites` heading.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
---------
Co-authored-by: Claude Opus 5 <noreply@anthropic.com>
## Summary
Fixes#663
- **Security fix**: `mcp-publisher login` no longer writes
`.mcpregistry_github_token` and `.mcpregistry_registry_token` to the
current working directory. GitHub tokens are kept in memory during the
login→exchange flow and cleared immediately after.
- **Token location**: Unified token moved from `~/.mcp_publisher_token`
to `~/.config/mcp-publisher/token.json` (dir 0700, file 0600), following
the convention used by `gh`, `docker`, and `kubectl`.
- **Dead code removed**: `NeedsLogin()` was on the `Provider` interface
but never called — removed from interface and all 4 implementations.
`forceLogin` field also removed.
- **Auth tests added**: 11 new tests for `GitHubATProvider` (package
previously had 0 tests).
- **Test isolation**: `SetupTestToken` now uses `t.TempDir()` +
`t.Setenv("HOME", ...)` instead of writing to real `$HOME`.
### Breaking change
Users must re-login after upgrading (`mcp-publisher login <method>`).
Existing tokens at `~/.mcp_publisher_token` are not auto-migrated.
`logout` cleans up both old and new locations.
## Test plan
- [x] `make check` passes (lint, unit tests, integration tests)
- [x] `mcp-publisher login none` stores token at
`~/.config/mcp-publisher/token.json`
- [x] No `.mcpregistry_*` files created in cwd after login (with
`--token`, `MCP_GITHUB_TOKEN`, or device flow)
- [x] `mcp-publisher publish` reads token from new location
- [x] `mcp-publisher validate` reads registry URL from new location
- [x] `mcp-publisher logout` cleans up new location, legacy
`~/.mcp_publisher_token`, and cwd `.mcpregistry_*` files
- [x] `mcp-publisher logout` when not logged in shows "Not logged in"
- [x] File permissions: dir 0700, file 0600
🤖 Generated with [Claude Code](https://claude.com/claude-code)
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
<!-- Provide a brief summary of your changes -->
## Motivation and Context
<!-- Why is this change needed? What problem does it solve? -->
The following PR leverages ko instead of docker build for building the
container image.
The motivation for the move is that ko has been one of the widely
adopted ways to build Go applications, but overall its benefits are:
* Build time is much faster - around 70-80%
* The resulting image size is much smaller (also more secure) - around
50% smaller
The changes in this PR are:
* Updated the workflows for staging and release to prod to use it
* Updated the local development + the integration tests to use it (the
integration tests overwrite the base image from chainguards one to
alpine mostly because we rely on wget to check the health status and you
don't get that in the stripped down one)
From the user point of view everything should continue to work as it is.
## How Has This Been Tested?
<!-- Have you tested this in a real application? Which scenarios were
tested? -->
## Breaking Changes
<!-- Will users need to update their code or configurations? -->
## Types of changes
<!-- What types of changes does your code introduce? Put an `x` in all
the boxes that apply: -->
- [ ] Bug fix (non-breaking change which fixes an issue)
- [ ] New feature (non-breaking change which adds functionality)
- [ ] Breaking change (fix or feature that would cause existing
functionality to change)
- [ ] Documentation update
## Checklist
<!-- Go over all the following points, and put an `x` in all the boxes
that apply. -->
- [ ] I have read the [MCP
Documentation](https://modelcontextprotocol.io)
- [ ] My code follows the repository's style guidelines
- [ ] New and existing tests pass locally
- [ ] I have added appropriate error handling
- [ ] I have added or updated documentation as needed
## Additional context
<!-- Add any other context, implementation notes, or design decisions
-->
---------
Signed-off-by: Radoslav Dimitrov <radoslav@stacklok.com>
<!-- Provide a brief summary of your changes -->
## Motivation and Context
<!-- Why is this change needed? What problem does it solve? -->
The following PR:
* Makes mcpb package version property optional
* Bumps the schema and reflects that in the code
* Makes sure the publisher tests are being covered by doing `make check`
## How Has This Been Tested?
<!-- Have you tested this in a real application? Which scenarios were
tested? -->
## Breaking Changes
<!-- Will users need to update their code or configurations? -->
## Types of changes
<!-- What types of changes does your code introduce? Put an `x` in all
the boxes that apply: -->
- [ ] Bug fix (non-breaking change which fixes an issue)
- [ ] New feature (non-breaking change which adds functionality)
- [ ] Breaking change (fix or feature that would cause existing
functionality to change)
- [ ] Documentation update
## Checklist
<!-- Go over all the following points, and put an `x` in all the boxes
that apply. -->
- [ ] I have read the [MCP
Documentation](https://modelcontextprotocol.io)
- [ ] My code follows the repository's style guidelines
- [ ] New and existing tests pass locally
- [ ] I have added appropriate error handling
- [ ] I have added or updated documentation as needed
## Additional context
<!-- Add any other context, implementation notes, or design decisions
-->
---------
Signed-off-by: Radoslav Dimitrov <radoslav@stacklok.com>
Remove redundant endpoint as suggested in #617. Clients should now use
`GET /v0/servers/{serverName}/versions/latest` instead (using `latest`
as a special version value in the existing endpoint).
This simplifies the API surface and reduces implementation burden for
subregistries, while maintaining a clear migration path for consumers.
## Changes
- Removed `GET /v0/servers/{serverName}` endpoint
- Updated `GET /v0/servers/{serverName}/versions/{version}` to accept
`latest` as special version
- Updated all tests to use new endpoint pattern
- Updated documentation and OpenAPI spec
- Added breaking change notice to CHANGELOG
Fixes#617
Co-authored-by: Claude <noreply@anthropic.com>
<!-- Provide a brief summary of your changes -->
## Motivation and Context
<!-- Why is this change needed? What problem does it solve? -->
Housekeeping on leftover psql references
## How Has This Been Tested?
<!-- Have you tested this in a real application? Which scenarios were
tested? -->
## Breaking Changes
<!-- Will users need to update their code or configurations? -->
## Types of changes
<!-- What types of changes does your code introduce? Put an `x` in all
the boxes that apply: -->
- [ ] Bug fix (non-breaking change which fixes an issue)
- [ ] New feature (non-breaking change which adds functionality)
- [ ] Breaking change (fix or feature that would cause existing
functionality to change)
- [ ] Documentation update
## Checklist
<!-- Go over all the following points, and put an `x` in all the boxes
that apply. -->
- [ ] I have read the [MCP
Documentation](https://modelcontextprotocol.io)
- [ ] My code follows the repository's style guidelines
- [ ] New and existing tests pass locally
- [ ] I have added appropriate error handling
- [ ] I have added or updated documentation as needed
## Additional context
<!-- Add any other context, implementation notes, or design decisions
-->
Signed-off-by: Radoslav Dimitrov <radoslav@stacklok.com>
<!-- Provide a brief summary of your changes -->
## Motivation and Context
<!-- Why is this change needed? What problem does it solve? -->
The follow PR removes the in-memory database option from the registry.
We had a chat about this with @domdomegg and although it is supposed to
be convenient, it isn't that useful in reality. We want to test and
develop the registry in a setup that is closer to the environment it is
supposed to be run. One of the main reasons are:
* We don't use it anywhere else besides unit testing and potentially
local development. Instead we should use postgresql for reliable results
(in-memory was actually giving false positive results)
* We have to keep maintaining it and make sure it mimics the behaviour
we have for postgresql which is not yielding a lot of value
## How Has This Been Tested?
<!-- Have you tested this in a real application? Which scenarios were
tested? -->
Yes, I've updated the test unit make target to run in the postgresql
environment and now we can see the race condition issue happening.
## Breaking Changes
<!-- Will users need to update their code or configurations? -->
No
## Types of changes
<!-- What types of changes does your code introduce? Put an `x` in all
the boxes that apply: -->
- [ ] Bug fix (non-breaking change which fixes an issue)
- [ ] New feature (non-breaking change which adds functionality)
- [ ] Breaking change (fix or feature that would cause existing
functionality to change)
- [ ] Documentation update
## Checklist
<!-- Go over all the following points, and put an `x` in all the boxes
that apply. -->
- [ ] I have read the [MCP
Documentation](https://modelcontextprotocol.io)
- [ ] My code follows the repository's style guidelines
- [ ] New and existing tests pass locally
- [ ] I have added appropriate error handling
- [ ] I have added or updated documentation as needed
## Additional context
<!-- Add any other context, implementation notes, or design decisions
-->
---------
Signed-off-by: Radoslav Dimitrov <radoslav@stacklok.com>
Wielded Claude Code to do this; across many iterations and have reviewed
all the code + largely rewritten the PR commentary here.
This PR implements the server ID consistency changes from
[#396](https://github.com/modelcontextprotocol/registry/issues/396). Our
goal is to:
- Rename `id` to `versionId` to make it clear that it is the ID for a
specific version, not a server
- Introduce `serverId` as a net new concept. This will be 1:1 with
`name`. While not strictly necessary (we could just use name as a PK),
the convention of having a UUID as an ID on our objects is just a
helpful subjective architecture decision. Plus also leaves open the door
to reconsidering the "should we allow renames" question.
## Key Changes
### 🗄️ Database Schema Changes
- **Renamed**: `id` column → `version_id` (unique identifier for each
server version)
### 📋 Metadata Structure Updates
- **JSON Schema**: Updated
`_meta.io.modelcontextprotocol.registry/official` to include:
- `serverId`: Consistent ID across all versions of a server
- `versionId`: Unique ID for this specific version (renamed from `id`)
### 🔧 API Endpoint Updates
- **New endpoint structure**: Changed from version-ID-based to
server-ID-based URLs:
- `GET /v0/servers/{server_id}` - Get latest version of server
- `GET /v0/servers/{server_id}?version=X.X.X` - Get specific version
- `GET /v0/servers/{server_id}/versions` - List all versions of a server
- `PUT /v0/servers/{server_id}?version=X.X.X` - Edit specific version
- **Consistent URL patterns**: All server endpoints now use `server_id`
as the primary identifier
### Database Migration Strategy
The migration (`005_add_server_id_rename_version_id.sql`):
1. Adds new `version_id` column
2. Generates deterministic UUIDs for servers with the same name
3. Updates JSON metadata to add `serverId` and rename `id` to
`versionId`
4. Copies existing `id` values to `version_id` column
5. Drops old `id` column and makes `version_id` the new primary key
## Testing
Besides of course automated tests passing, I:
1. Set up my local db with the pre-005-migration state, imported all the
production data
2. Ran the migration
3. Spot checked some diffs of the data that resulted --
```json
// before
{
"name": "io.github.ruvnet/ruv-swarm",
"_meta": {
"io.modelcontextprotocol.registry/official": {
"id": "0a9cda7f-2b73-4839-844b-6582f335de42",
"is_latest": false,
"updated_at": "2025-09-10T17:21:34.92137012Z",
"published_at": "2025-09-10T17:15:32.69690433Z"
}
},
"version": "1.0.18",
"repository": {
"url": "",
"source": ""
},
"description": "Neural network swarm orchestration with WebAssembly acceleration and MCP integration"
}
// after
{
"name": "io.github.ruvnet/ruv-swarm",
"_meta": {
"io.modelcontextprotocol.registry/official": {
"serverId": "e8538952-96af-4a33-9e31-7fbed012521e",
"is_latest": false,
"versionId": "0a9cda7f-2b73-4839-844b-6582f335de42",
"updated_at": "2025-09-10T17:21:34.92137012Z",
"published_at": "2025-09-10T17:15:32.69690433Z"
}
},
"version": "1.0.18",
"repository": {
"url": "",
"source": ""
},
"description": "Neural network swarm orchestration with WebAssembly acceleration and MCP integration"
}
```
Servers that have multiple versions stayed consistent in their
`serverId`:
```json
// First version
{
"name": "app.getdialer/dialer",
"_meta": {
"io.modelcontextprotocol.registry/official": {
"serverId": "80f08bdb-f3e4-4f19-b072-1676ae49b663",
"is_latest": false,
"versionId": "04f25b17-331d-48d7-b08b-37baece208a7",
"updated_at": "2025-09-09T00:16:49.162620072Z",
"published_at": "2025-09-08T23:47:09.452137971Z"
}
},
"status": "active",
"$schema": "https://static.modelcontextprotocol.io/schemas/2025-07-09/server.schema.json",
"remotes": [
{
"url": "https://getdialer.app/mcp",
"type": "streamable-http"
},
{
"url": "https://getdialer.app/sse",
"type": "sse"
}
],
"version": "1.0.0",
"repository": {
"url": "",
"source": ""
},
"description": "An MCP server that provides your you make outbound phone calls using your own phone number"
}
// Second version
{
"name": "app.getdialer/dialer",
"_meta": {
"io.modelcontextprotocol.registry/official": {
"serverId": "80f08bdb-f3e4-4f19-b072-1676ae49b663",
"is_latest": true,
"versionId": "faa26cb9-3aad-415e-aece-f715fff5fe5e",
"updated_at": "2025-09-09T00:16:49.107842808Z",
"published_at": "2025-09-09T00:16:49.107842808Z"
}
},
"status": "active",
"$schema": "https://static.modelcontextprotocol.io/schemas/2025-07-09/server.schema.json",
"remotes": [
{
"url": "https://getdialer.app/mcp",
"type": "streamable-http"
},
{
"url": "https://getdialer.app/sse",
"type": "sse"
}
],
"version": "1.0.1",
"repository": {
"url": "",
"source": ""
},
"description": "An MCP server that provides lets you make outbound phone calls using your own phone number"
}
```
And I also tried publishing a test server anew; looked good:
```bash
⏺ Bash(curl -s "http://localhost:8080/v0/servers/102dc4e4-6ca8-40e6-aa11-cd45763d3352" | jq '.')
⎿ {
"$schema": "https://static.modelcontextprotocol.io/schemas/2025-07-09/server.schema.json",
"name": "io.modelcontextprotocol.anonymous/test-claude-migration-server",
"description": "A test server to verify the new migration and API endpoints work correctly",
"status": "active",
"repository": {
"url": "https://github.com/example/test-claude-migration-server",
"source": "github",
"id": "example/test-claude-migration-server"
},
"version": "1.0.0",
"_meta": {
"io.modelcontextprotocol.registry/official": {
"serverId": "102dc4e4-6ca8-40e6-aa11-cd45763d3352",
"versionId": "48c9cd55-7bc4-4ac2-896e-59e81eb653e4",
"published_at": "2025-09-12T18:33:23.264203-07:00",
"updated_at": "2025-09-12T18:33:23.264203-07:00",
"is_latest": true
}
}
}
⏺ Bash(curl -s "http://localhost:8080/v0/servers/102dc4e4-6ca8-40e6-aa11-cd45763d3352/versions" | jq '.')
⎿ {
"servers": [
{
"$schema": "https://static.modelcontextprotocol.io/schemas/2025-07-09/server.schema.json",
"name": "io.modelcontextprotocol.anonymous/test-claude-migration-server",
"description": "A test server to verify the new migration and API endpoints work correctly",
"status": "active",
"repository": {
"url": "https://github.com/example/test-claude-migration-server",
"source": "github",
"id": "example/test-claude-migration-server"
},
"version": "1.0.0",
"_meta": {
"io.modelcontextprotocol.registry/official": {
"serverId": "102dc4e4-6ca8-40e6-aa11-cd45763d3352",
"versionId": "48c9cd55-7bc4-4ac2-896e-59e81eb653e4",
"published_at": "2025-09-12T18:33:23.264203-07:00",
"updated_at": "2025-09-12T18:33:23.264203-07:00",
"is_latest": true
}
}
}
],
"metadata": {
"count": 1
}
}
```
Co-Authored-By: Claude <noreply@anthropic.com>
---------
Co-authored-by: Claude <noreply@anthropic.com>
Co-authored-by: Preeti Dewani <preetidewani1990@gmail.com>
Co-authored-by: adam jones <adamj+git@anthropic.com>
Co-authored-by: Adam Jones <adamj@anthropic.com>
## Summary
Simplified the schema structure by:
- Replacing `version_detail.version` with direct `version` field
- Removing unused `release_date` field from seed data
- Updating all API types, models, and tests to match new structure
- Maintaining backwards compatibility in API responses
This change makes the schema more intuitive and reduces nesting depth.
## Summary
Updates the `_meta` field structure to be compliant with the MCP
specification by using proper `prefix/name` format for vendor
extensions.
### Changes
- **Registry metadata**: `io.modelcontextprotocol.registry` →
`io.modelcontextprotocol.registry/official`
- **Publisher metadata**: `publisher` →
`io.modelcontextprotocol.registry/publisher-provided`
- **Database migration**: Updates indexes to use new field paths
- **Updated all code**: Service layer, validators, tests, and seed data
### Format Before
```json
{
"_meta": {
"publisher": { "license": "MIT" },
"io.modelcontextprotocol.registry": { "id": "...", "is_latest": true }
}
}
```
### Format After
```json
{
"_meta": {
"io.modelcontextprotocol.registry/publisher-provided": { "license": "MIT" },
"io.modelcontextprotocol.registry/official": { "id": "...", "is_latest": true }
}
}
```
This resolves the naming collision with the planned top-level
`publisher` field and ensures compliance with MCP specification
requirements.
Fixes#356
## Motivation and Context
Docs currently don't have very clear organisation, and don't explain
core flows.
This fixes#89 (although we still might want to push these to
`https://modelcontextprotocol.io/docs/`), based on the plan there.
One minor change from that plan is that I merged tutorials and guides,
because they ended up basically being duplicates which I thought would
be annoying for us to keep in sync. Additionally in the Diátaxis
framework, tutorials and guides are separated by whether they are
focused on acquisition of skills (e.g. in a learning environment) or
application of skills (e.g. in a work environment). I expect the
registry to be something that will be used occasionally in a work
environment, rather than a fundamental 'learnable' skill so I think it's
reasonable not to have tutorials and fold these into guides.
## How Has This Been Tested?
- Read through all these updated docs
- Unit and integration tests (that previously used examples.md and use
the schema) working
- Ran `npx markdown-link-check .` to check all the links work
## Breaking Changes
Links to old docs will break, but I don't think there are any. All
internal links have been checked and updated.
## Types of changes
<!-- What types of changes does your code introduce? Put an `x` in all
the boxes that apply: -->
- [ ] Bug fix (non-breaking change which fixes an issue)
- [ ] New feature (non-breaking change which adds functionality)
- [ ] Breaking change (fix or feature that would cause existing
functionality to change)
- [x] Documentation update
Fixes#96
## Motivation and Context
See #96
## How Has This Been Tested?
Has some unit and integration tests, still needs proper testing of the
different registry types
## Breaking Changes
Yes, new package uploads will need extra verification
## Types of changes
<!-- What types of changes does your code introduce? Put an `x` in all
the boxes that apply: -->
- [ ] Bug fix (non-breaking change which fixes an issue)
- [x] New feature (non-breaking change which adds functionality)
- [x] Breaking change (fix or feature that would cause existing
functionality to change)
- [ ] Documentation update
## Motivation and Context
The database logic was getting a bit unwieldy and fragile - especially
because business logic that should really have lived at the registry
service level was being pushed into the persistence level.
This PR tightens the interface, simplifies the code there, as well as
rips out a bunch of mocks etc. we don't need.
## How Has This Been Tested?
Locally everything works
## Breaking Changes
The list endpoint was surprisingly always filtering for isLatest only,
which seems like a bug? I have removed this filter.
## Summary
Move the publisher CLI from `tools/publisher/` to `cmd/publisher/` as
it's a proper application rather than a development utility.
This addresses feedback from @rdimitrov in PR #324 that the publisher
CLI should be under `cmd/` since it's not really a tool in the same
sense as other helpers in the `tools/` folder.
## Changes
- Moved `tools/publisher/` directory to `cmd/publisher/`
- Updated Makefile build and clean targets
- Updated README.md documentation links
- Updated VS Code launch configuration
- Updated integration test script
- Updated FAQ documentation
- Fixed Go import paths in publisher code
---------
Co-authored-by: claude[bot] <209825114+claude[bot]@users.noreply.github.com>
Co-authored-by: adam jones <domdomegg@users.noreply.github.com>
## Summary
- Improved command structure and organization for the MCP Publisher CLI
- Enhanced readability and maintainability of publisher commands
- Cleaned up publish and edit handler logic
This builds on the previous cleanup work to further streamline the
publisher interface.
This also works as part of
https://github.com/modelcontextprotocol/registry/issues/89
<!-- Provide a brief summary of your changes -->
This PR serves as a placeholder and brings together all of the changes
related to supporting vendor extensions passed through the API. The idea
is to have a single place to track and review the work as it comes
together (even if it’s still a bit of a work in progress).
**The changes in this PR cover:**
* Updates on the data model
* Updates on both databases (in-memory and postgresql), added a new
table to PSQL for tracking the extensions leaving the server table only
for the clean server.json properties
* Updates the seed.json file (thus the big line change)
* Adds a seed-migrate tool to help migrating an existing seed file to
the new format
* Updates the seed import mechanism to support this
* Updates the existing tests - unit, integration and validation tests
* Adds the publish handler to the OpenAPI spec (it was missing)
* Fixes linting errors
* Updates on the API side on (publish, list, get) to support the new
format of `ServerResponse` -
```
{
"server": {},
"x-io.modelcontextprotocol.registry": {}, /*not available for publish since this gets populated by the registry*/
"x-publisher": {}
}
```
## Motivation and Context
<!-- Why is this change needed? What problem does it solve? -->
Motivated by the discussions in:
https://github.com/modelcontextprotocol/registry/discussions/284
## How Has This Been Tested?
<!-- Have you tested this in a real application? Which scenarios were
tested? -->
Locally and by running all of the tests
## Breaking Changes
<!-- Will users need to update their code or configurations? -->
Yes, it changes both the API and the format of the seed file
## Types of changes
<!-- What types of changes does your code introduce? Put an `x` in all
the boxes that apply: -->
- [ ] Bug fix (non-breaking change which fixes an issue)
- [ ] New feature (non-breaking change which adds functionality)
- [x] Breaking change (fix or feature that would cause existing
functionality to change)
- [ ] Documentation update
---
Signed-off-by: Radoslav Dimitrov <radoslav@stacklok.com>
Co-authored-by: adam jones <domdomegg+git@gmail.com>
Fixes#228
## Motivation and Context
See #19
## How Has This Been Tested?
- Run locally
- Run in docker compose
- Run infra in minikube
- Unit and integration tests all passing :)
## Breaking Changes
Yes: users will not be able to access any data they used to have in
custom mongodb. We are accepting this breaking change, given the
registry is in experimental development status.
## Types of changes
<!-- What types of changes does your code introduce? Put an `x` in all
the boxes that apply: -->
- [ ] Bug fix (non-breaking change which fixes an issue)
- [ ] New feature (non-breaking change which adds functionality)
- [x] Breaking change (fix or feature that would cause existing
functionality to change)
- [ ] Documentation update
## Checklist
<!-- Go over all the following points, and put an `x` in all the boxes
that apply. -->
- [x] I have read the [MCP
Documentation](https://modelcontextprotocol.io)
- [x] My code follows the repository's style guidelines
- [x] New and existing tests pass locally
- [x] I have added appropriate error handling
- [x] I have added or updated documentation as needed
## Additional context
I have not implemented connection pooling support for simplicity, and
because cloudnative-pg already provides built-in connection pooling with
pgBouncer so if we want it later we can switch to this endpoint easily.
## Motivation and Context
- Adds credentials for the new MCP Registry Login GitHub Apps (local,
staging, prod)
- Adds support for "none" auth to publisher
## How Has This Been Tested?
- Tested publisher workflows locally
- Unit and integration tests passing
## Breaking Changes
None
## Types of changes
<!-- What types of changes does your code introduce? Put an `x` in all
the boxes that apply: -->
- [ ] Bug fix (non-breaking change which fixes an issue)
- [x] New feature (non-breaking change which adds functionality)
- [ ] Breaking change (fix or feature that would cause existing
functionality to change)
- [ ] Documentation update
## Checklist
<!-- Go over all the following points, and put an `x` in all the boxes
that apply. -->
- [x] I have read the [MCP
Documentation](https://modelcontextprotocol.io)
- [x] My code follows the repository's style guidelines
- [x] New and existing tests pass locally
- [x] I have added appropriate error handling
- [x] I have added or updated documentation as needed
---
Merging based on Tadas's approval review, and only making minor docs changes on top of that.
Fixes#264. Adds endpoints:
```
/v0/auth/github # exchange a github token for a registry token that gives you access to namespaces io.github.username and io.github.org for your orgs
/v0/auth/none # get a registry token for namespace io.modelcontextprotocol.anonymous (only for testing/local dev purposes)
```
Updates `/v0/publish` to accept one of these registry tokens.
## Motivation and Context
See #264
## How Has This Been Tested?
- Running locally and getting a token from `/v0/auth/none` and then
publishing to `/v0/publish`
- Updating the integration tests that get a token and then publish, and
them passing
## Breaking Changes
This is a breaking change to the auth API. This is fine for registry
given it's not deployed yet.
## Types of changes
<!-- What types of changes does your code introduce? Put an `x` in all
the boxes that apply: -->
- [ ] Bug fix (non-breaking change which fixes an issue)
- [x] New feature (non-breaking change which adds functionality)
- [x] Breaking change (fix or feature that would cause existing
functionality to change)
- [x] Documentation update
## Checklist
<!-- Go over all the following points, and put an `x` in all the boxes
that apply. -->
- [x] I have read the [MCP
Documentation](https://modelcontextprotocol.io)
- [x] My code follows the repository's style guidelines
- [x] New and existing tests pass locally
- [x] I have added appropriate error handling
- [x] I have added or updated documentation as needed
## Additional context
<!-- Add any other context, implementation notes, or design decisions
-->
---------
Co-authored-by: claude[bot] <209825114+claude[bot]@users.noreply.github.com>
Co-authored-by: adam jones <domdomegg@users.noreply.github.com>
This PR simplifies the seed data import configuration based on
maintainer feedback by removing command line flags and using a clearer
environment variable name.
## Key Changes
### Configuration Simplification
- **Renamed**: `MCP_REGISTRY_SEED_FILE_PATH` → `MCP_REGISTRY_SEED_FROM`
- **Removed**: `--seed-file-path` command line flag to eliminate
configuration complexity
### Improved User Experience
Users now configure seed data import using a single, clear environment
variable:
```bash
# Before (multiple ways to configure)
MCP_REGISTRY_SEED_FILE_PATH=data/seed.json ./registry
./registry --seed-file-path=data/seed.json
# After (simplified single method)
MCP_REGISTRY_SEED_FROM=data/seed.json ./registry
```
### Registry Composability
The composability workflow is now simplified while maintaining all
functionality:
```bash
# Start first registry with seed data
MCP_REGISTRY_SEED_FROM=data/seed.json ./registry
# Start second registry importing from first via pagination
MCP_REGISTRY_SEED_FROM=http://localhost:8080 ./registry
```
## Technical Implementation
- Maintains full backward compatibility for import functionality
- All existing seed data sources continue to work (local files, HTTP
URLs, registry URLs)
- Automatic pagination through `/v0/servers` endpoint remains unchanged
- Configurable delays between requests preserved
## Benefits
- **Reduced complexity**: Single configuration method instead of
multiple
- **Clearer naming**: `SEED_FROM` better describes the source of seed
data
- **Easier maintenance**: Fewer configuration paths to test and document
- **Consistent patterns**: Aligns with other environment-variable-only
configuration
All existing functionality is preserved while providing a cleaner, more
maintainable configuration interface.
<!-- START COPILOT CODING AGENT TIPS -->
---
💬 Share your feedback on Copilot coding agent for the chance to win a
$200 gift card! Click
[here](https://survey.alchemer.com/s3/8343779/Copilot-Coding-agent) to
start the survey.
---------
Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
Co-authored-by: Avinash Sridhar <sridharavinash@users.noreply.github.com>
Co-authored-by: Toby Padilla <toby@toby.sh>
Co-authored-by: toby <83556+toby@users.noreply.github.com>
