django-xprez-mcp
Admin-style MCP servers for Django. Register providers, models, admin classes and plain functions on a server; they become MCP tools, resources and prompts, filtered by the calling user's permissions.
Early development: phases 1 (core), 2 (models) and 3 (admin). See docs/design.md for the full design and roadmap.
# settings.py
INSTALLED_APPS = [..., "xprez_mcp", "xprez_mcp.contrib.tokens"]
XPREZ_MCP = {
"PERMISSION": "myapp.use_mcp", # required: who may use the server at all
}
# urls.py
from xprez_mcp import mcp
urlpatterns = [path("mcp/", mcp.urls)]
Clients send Authorization: Bearer <key>. With xprez_mcp.contrib.tokens
users create keys under "MCP tokens" in the admin (shown once, stored hashed,
revoked by deleting). Projects with their own token model leave the app out
and set XPREZ_MCP["TOKEN_RESOLVER"] = "myapp.tokens.resolve"
((request, token) -> user | None).
# myapp/mcp.py: discovered automatically, like admin.py
from xprez_mcp import Provider, mcp, prompt, resource, tool
@mcp.register
class NotesProvider(Provider):
namespace = "notes"
@tool(read_only=True, permission="myapp.view_note")
def search(self, ctx, query: str, limit: int = 20):
"""Search notes by title."""
notes = Note.objects.filter(owner=ctx.user, title__icontains=query)
return {"notes": list(notes.values("id", "title")[:limit])}
@resource("notes://{id}", mime_type="text/markdown")
def note(self, ctx, id: int):
return Note.objects.get(owner=ctx.user, pk=id).text
@prompt
def summary(self, ctx, id: int):
"""Summarize one note."""
return f"Summarize the note notes://{id}."
@mcp.tool
def ping(ctx) -> str:
return "pong"
# myapp/mcp.py: tools generated from a model
from xprez_mcp import ModelProvider, mcp
@mcp.register(Note)
class NoteProvider(ModelProvider):
fields = ["id", "title", "created"]
search_fields = ["title"]
filters = {"created_after": "created__gte"}
actions = ["list", "get", "update"] # notes_list, notes_get, notes_update
def get_queryset(self, ctx):
return Note.objects.filter(owner=ctx.user)
With many models, MCPServer(models="generic") exposes them through a few
shared tools (models_list, records_search, record_get, ...) instead of
tools per model. A provider can choose for itself with exposure.
# the admin, with exactly the permissions each user has there
from django.contrib import admin
from xprez_mcp import AdminProvider, mcp
@mcp.register(Note)
class NoteAdminProvider(AdminProvider): # uses the registered NoteAdmin
actions = ["list", "get", "update"] # writes are opt-in
admin_actions = ["publish"] # ModelAdmin actions, opt-in
# the rest of admin.site (models registered above keep their own provider)
mcp.register_admin(admin.site, exposure="generic")
Write tools from installed third-party apps are held back until listed in
XPREZ_MCP["ALLOW_WRITES"].
Development
python3 -m venv .venv
.venv/bin/pip install -e ".[dev]"
.venv/bin/python -m django test xprez_mcp.tests --settings=xprez_mcp.tests.settings
.venv/bin/black --check .
Metadata
Release files for django-xprez-mcp 0.4.2
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| django_xprez_mcp-0.4.2.tar.gz | 83.4 kB | With provenance | Details for django_xprez_mcp-0.4.2.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| django_xprez_mcp-0.4.2-py3-none-any.whl | Python 3 | none | any | Details for django_xprez_mcp-0.4.2-py3-none-any.whl |
Total release size: 167.7 kB
Release files / django_xprez_mcp-0.4.2.tar.gz
| Download URL | django_xprez_mcp-0.4.2.tar.gz |
|---|---|
| Size | 83.4 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
d6cfbe83d5ac4466720c70ca72e2dbd5cf54c5bcf4664d3ae54d1f0d67a7f4cf
|
|
BLAKE2b-256 checksum How to use checksums |
69724bf879f3dbc19c3412079268381b24035331f04d64e38d27655ae27d14e5
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.16
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Nov 19, 2026.
Transparency logRelease files / django_xprez_mcp-0.4.2-py3-none-any.whl
| Download URL | django_xprez_mcp-0.4.2-py3-none-any.whl |
|---|---|
| Size | 84.3 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
f02073465818b9552fb9267153ca31c32e7215a992ebfcd044f395d33fe04f91
|
|
BLAKE2b-256 checksum How to use checksums |
b581873dfe3926f37a83439cbd193c34448fcc6f0ce76e58e3d0e821666cd274
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.16
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Nov 19, 2026.
Transparency logSecurity for django-xprez-mcp 0.4.2
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
Full provenance
Both files in this release have been uploaded with provenance metadata.
Attestation for django_xprez_mcp-0.4.2.tar.gz
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Nov 19, 2026.
Transparency logA note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.