Society Gate MCP Server
Provides an Alexa+ MCP server and classic Alexa custom skill flow for society-gate visitor approval, including voice confirmation, an Alexa request/response handler, an en-IN interaction model, and Lambda relay support.
This server does not currently have a deployable release on Glama. Follow the repository's setup instructions to install and configure it for your MCP client, including any required credentials and local dependencies.
Society Gate
A separate, local prototype for extending Alexa from the home to the society gate.
For the Alexa+ experience, start the real Streamable HTTP MCP server and use the browser's MCP panel: setup and demo. It negotiates protocol 2025-11-25, discovers five tools, and commits decisions only after a resident-bound confirmation. The classic Alexa skill simulator below remains available separately.
Visitor arrives → guard records arrival → resident hears the request through Alexa → resident confirms → guard receives the decision. A resident can also create a time-limited rule for a regular visitor with a verified vendor identity.
You said the society already has an app, but you do not have its API access. This build therefore works as a companion system. It does not read, modify, or synchronize that app. The integration endpoint and mock vendor account let us develop and test the connection contract until the vendor provides access.
Run locally
Python 3.11+; no installation, Docker, cloud account, or paid service needed for the base companion. Run from this application's directory:
python3 -m society_gate.serverOpen http://127.0.0.1:8091. In a second terminal, read the generated demo credentials (keep these local):
cat .local/credentials.jsonPaste the guard token into one browser tab and the resident token into
another. The resident is enrolled in A-101; neighbour is enrolled in B-202.
Tokens stay in the current browser tab's session storage until disconnect.
All visitors, audit entries, rules, and voice sessions persist in SQLite across
server restarts. .local/ is ignored by Git.
For a second isolated demo, run python3 -m society_gate.server --port 8092 --data-dir /tmp/society-gate-demo as one command. The demo binds only to loopback.
Related MCP server: ADOS MCP
Try the full flow
Guard tab: register Rahul for A-101, purpose Delivery.
Resident tab: click Open skill, Approve visitor, then Yes, confirm. These buttons exercise the Alexa conversation handler without an Echo or microphone. The resident can also approve or deny directly on the dashboard.
Guard tab: the request updates to Approved within 3 seconds. Click Record entry, then Record exit when the visitor leaves.
Review the activity log. Every transition identifies its actor.
Unanswered requests expire after five minutes. An approval is valid for two minutes. A timeout never grants entry. The guard still checks visitor identity and physically manages entry; this project does not operate a gate motor.
Try a regular-visitor rule
Resident tab: add a rule labelled Housekeeper for vendor ID
vendor:regular-001, valid for one hour.Open a third tab with the
integrationtoken. This is a mock vendor, not a connection to your existing app.Submit an arrival for A-101 using that ID and check the mock identity-verified box. The resident's active rule approves the arrival automatically.
Repeat with an unverified identity or a different ID: it requires approval. A guard-entered name or reference never activates an automatic rule.
Revoke the rule from the resident tab: future arrivals need approval and any unused approval from that rule is withdrawn. Already recorded entries remain in the log.
Rules last at most 24 hours and apply only to the resident's flat. The mock vendor's verified flag is a test assertion; a real adapter must derive it from the vendor's documented identity verification, never visitor-supplied data.
What is implemented
Responsive guard/resident dashboard and mock vendor workspace.
Persistent requests, rules, actor identities, audit records, and voice sessions.
Flat-scoped decisions, explicit voice confirmation, expiration, and single-use entry.
Transactional decisions and deduplication for vendor events and Alexa retries.
Alexa request/response handler, en-IN interaction model, and Lambda relay code.
Alexa+ web simulation using a real MCP client and server, with protocol trace.
A generic authenticated inbound integration contract.
Automated domain and HTTP tests, using Python's standard library.
What still needs external setup
A real Echo is not connected yet. See Alexa setup. It requires an Amazon developer skill, an IAM-restricted Lambda trigger, a deployed HTTPS backend, and OAuth account linking to enrolled residents. The adapter accepts linked access tokens; this prototype does not implement a production OAuth authorization server or account enrollment service.
Those steps are for the optional classic Alexa custom skill. The Alexa+ MCP simulation runs locally without preview access or a physical device; see MCP setup for its separate integration limits.
Your existing society app is not connected. See the integration contract. We need vendor-supported inbound arrival events, household mappings, stable verified visitor IDs, and an outbound decision API. The fallback available today is the companion guard dashboard.
The local version does not send phone push alerts or unsolicited Alexa speech. Residents open the skill or dashboard to see pending requests. Alexa proactive events require opt-in and predefined event schemas; we have not established a suitable visitor-approval notification type. See Amazon's proactive events documentation.
Verification
python3 -m unittest discover -s tests -vThis suite tests the state machine, cross-flat authorization, pending/approval timeouts, concurrent decisions, durable records, explicit voice confirmation, session boundaries, replay handling, verified-ID rules, revocation, HTTP authorization, and the Alexa relay boundary. It does not test a physical Echo, Amazon certification, or a society vendor integration.
The code in this standalone project is MIT licensed. To export a source archive
without credentials or local state, run python3 scripts/package_source.py.
Read the product assessment for the integration scope and the architecture decision for the application design.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Related MCP Connectors
Tenant-scoped credentials, training and business records with owner-controlled write approvals.
Governance runtime for compliance: verified, human-approved writes to a tamper-evident record.
Buzz in deliveries and guests: apartment access codes, timed or recurring entry, and access history.
Controls for your own verified API: read events, block or release callers, tune route limits.
Related MCP Servers
- FlicenseNot gradedqualityCmaintenanceExposes the Veridexa S2S API to MCP clients (e.g., ChatGPT) via Streamable HTTP, allowing document verification and report retrieval.-
- AlicenseNot gradedqualityBmaintenanceEnables AI clients to read live drone status, telemetry, and parameters, and to control drone operations with scoped permissions, confirmation steps, and audit trails.MIT
- FlicenseNot gradedqualityBmaintenanceEnables managing projects, features, tasks, dependencies, executions, and human review through a Streamable HTTP endpoint, along with persistent inter-agent cooperation via task messaging and event subscriptions. It also supports administrative actions such as approvals, cancellations, unblocking, and member management.-
- AlicenseAqualityCmaintenanceEnables AI agents to browse customers, bookings, and availability while any booking creation or cancellation stays gated behind a human-approved, single-use token cryptographically bound to the exact arguments. Runs over stdio or Streamable HTTP and enforces the read/write separation in host code with a full audit trail.7MIT