Slide

Slide MCP Server

MCP server for Slide's backup/BDR (business continuity/disaster recovery) API - account, agent, device, network, backup, snapshot, and restore visibility, for AI assistants and the WYRE Conduit gateway.

Authentication

Slide authenticates with a static bearer API token, issued per-account in the Slide Console (Settings > API Keys) - not OAuth. This connector never mints or refreshes anything: it only ever holds a live token, sent as Authorization: Bearer <token> to api.slide.tech. In gateway mode the token arrives per-request via the X-Slide-Api-Token header and a request missing that header is rejected. In local/stdio mode, and in HTTP AUTH_MODE=env, it's read from SLIDE_API_TOKEN.

Configuration

Copy .env.example and set the variables below. Leave CONDUIT_S2S_SECRET empty in that file.

/health and /healthz stay unauthenticated and do not read credentials. The stdio transport does not use CONDUIT_S2S_SECRET.

HTTP authentication

With MCP_TRANSPORT=http and CONDUIT_S2S_SECRET set, every /mcp request must include X-Gateway-S2S. The check is verifyS2sHeader in src/s2s-verify.ts: the header value is t=<unix seconds>,v1=<hex>, where the hex is HMAC-SHA256 of the literal string t=<unix seconds> keyed with CONDUIT_S2S_SECRET (64 lowercase hex characters). The timestamp must be within 300 seconds of the server clock. A missing or invalid header is HTTP 401, before the Slide token is read.

TS=$(date +%s)SIG=$(printf 't=%s' "$TS" | openssl dgst -sha256 -hmac "$CONDUIT_S2S_SECRET" -hex | awk '{print $NF}')curl -sS http://127.0.0.1:8080/mcp \  -H 'Content-Type: application/json' \  -H 'Accept: application/json, text/event-stream' \  -H "X-Gateway-S2S: t=${TS},v1=${SIG}" \  -H "X-Slide-Api-Token: $SLIDE_API_TOKEN" \  -d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'

In AUTH_MODE=gateway, /mcp also requires X-Slide-Api-Token. That header is the only credential source for the request. A missing header is HTTP 401 and does not fall back to SLIDE_API_TOKEN.

AUTH_MODE=env still requires a valid X-Gateway-S2S header, then uses SLIDE_API_TOKEN when the request does not carry X-Slide-Api-Token.

Request limits

• HTTP request bodies are capped at 256 KiB (262144 bytes). After the X-Gateway-S2S check and (in AUTH_MODE=gateway) the X-Slide-Api-Token presence check pass, a Content-Length above the cap returns 413 with a JSON-RPC error; bodies without Content-Length are capped by the MCP SDK transport. Unauthenticated requests still get 401 first.
• tools/call arguments are capped at 10,000 elements (every array item and object key at any depth). Larger inputs return an isError tool result and are never dispatched.

These limits address CVE-2026-111768 / GHSA-46jr-4wx8-v453 (memory exhaustion during validation) and GHSA-f897-62hf-6gxq (no request body / batch cap).

Tools

Accounts

• slide_list_accounts - list accounts, filterable by creation date.
• slide_get_account - get a single account by ID.

Agents

• slide_list_agents - list agents (protected systems), filterable by device, client, type, and pairing date.
• slide_get_agent - get a single agent by ID (backup schedule, retention, volumes, alert configs).
• slide_search_agent_files - search the indexed file paths on an agent's protected system.
• slide_get_agent_file_versions - get the snapshot versions available for a file path on an agent.
• slide_get_agent_services - get the backed-up services detected on an agent.

Alerts

• slide_list_alerts - list alerts, filterable by device, agent, resolved status, and creation date.
• slide_get_alert - get a single alert by ID.

Audits

• slide_list_audit_logs - list audit log entries, filterable by action, resource type, client, and time range.
• slide_list_audit_actions - list the full set of audit action names.
• slide_list_audit_resource_types - list the full set of audit resource type names.
• slide_get_audit_log - get a single audit log entry by ID.

Backups

• slide_list_backups - list backup attempts, filterable by agent, snapshot, device, and time range.
• slide_get_backup - get a single backup attempt by ID.

Billing

• slide_list_billing_credit_memos - list billing credit memos.
• slide_list_billing_invoices - list billing invoices.
• slide_list_billing_subscriptions - list subscriptions and device licenses.

Clients

• slide_list_clients - list clients (the MSP's end customers), filterable by creation date.
• slide_get_client - get a single client by ID.

Devices

• slide_list_devices - list devices (physical/virtual Slide Boxes), filterable by client and creation date.
• slide_get_device - get a single device by ID.
• slide_get_device_network - get a device's network configuration.
• slide_list_device_vlans - list a device's configured VLANs.
• slide_get_device_vlan - get a single VLAN on a device by ID.

Networks

• slide_list_networks - list disaster-recovery networks, filterable by client and creation date.
• slide_get_network - get a single network by ID (port forwards, IPsec connections, WireGuard peers).

Restores (File)

• slide_list_file_restores - list file restores, filterable by creation date.
• slide_get_file_restore - get a single file restore by ID.
• slide_browse_file_restore - browse the files/folders exposed by a file restore.
• slide_list_file_restore_pushes - list push operations for a file restore, filterable by time range.
• slide_get_file_restore_push - get the status of a single file restore push.

Restores (Image)

• slide_list_image_exports - list image exports, filterable by creation date.
• slide_get_image_export - get a single image export by ID.
• slide_browse_image_export - browse the files/folders exposed by an image export.

Restores (Virtual Machine)

• slide_list_virtual_machines - list virtual machines created from a snapshot, filterable by creation date.
• slide_get_virtual_machine - get a single virtual machine by ID.

Snapshots

• slide_list_snapshots - list snapshots, filterable by agent, location/deletion state, and time range.
• slide_get_snapshot - get a single snapshot by ID.
• slide_get_snapshot_service_verification - get service-verification results for a snapshot.

Users

• slide_list_users - list users, filterable by creation date.
• slide_get_user - get a single user by ID.
• slide_get_user_avatar - get a user's avatar image (base64 data URL).

Scope

This is a deliberately read-only, non-credential-exposing v1 surface, covering all 43 of Slide's non-secret-exposing GET endpoints. Slide is backup/DR infrastructure - several of its real endpoints return live secrets, and every mutating endpoint is out of scope for v1. None of the excluded material is implemented here, by design, not by oversight. Every tool is classified isAdmin: true in the Conduit wiring, given the sensitivity of backup/DR infrastructure data (client environments, device inventories, network topology).

Secret-bearing fields - stripped from an otherwise-useful response, never implemented as a pass-through:

• Agent.passphrases[].passphrase (the live agent-encryption passphrase) - stripped from slide_list_agents / slide_get_agent. The passphrase's name and agent_passphrase_id are kept.
• Network.ipsec_conns[].psk (the live IPsec pre-shared key) - stripped from slide_list_networks / slide_get_network.
• Network.wg_peers[].wg_private_key (the live WireGuard private key) - stripped from the same two tools. wg_public_key is not secret and is kept.
• VirtualMachine.vnc_password (the live VNC password) - stripped from slide_list_virtual_machines / slide_get_virtual_machine. The vnc connection-info array (host/port/websocket URI) and vnc_enabled are not secret and are kept.
• ImageExport.password (the live Samba password for the exported share) - stripped from slide_list_image_exports / slide_get_image_export. This field isn't one of Slide's headline "secret" endpoints, but it's the same class of live credential material as the four above, so it's held to the same bar.

Hard-excluded (mutating) - every POST/PATCH/PUT/DELETE endpoint, never implemented: creating/updating clients, networks, devices, agents, VLANs; starting backups; creating file/image/VM restores and push operations; managing IPsec connections, port forwards, and WireGuard peers; agent passphrase add/delete; device reboot/poweroff; and all other writes across every tag. This is real destructive/mutating infrastructure (delete client, delete network, delete restore, reboot device) - out of scope for v1 by design.

Out of v1 scope (not credential/provisioning, just not part of this connector's structured-data surface):

• GET /v1/billing/credit-memo/{id}/pdf (BillingCreditMemoPDF) - returns raw application/pdf binary, not structured data.
• GET /v1/billing/invoice/{id}/pdf (BillingInvoicePDF) - same.

They can be added as a follow-up if there's demand, after a deliberate scope decision - not by default.

Credential scope

This connector's own code is read-only by construction — but Slide also offers a real way to make the underlying credential read-only too, and we recommend using it.

Slide documents six account roles: Account Owner, Account Admin, Business Continuity Technician, Billing, and Read Only (read-only access to all resources). An API token inherits the role of whoever generates it in the Slide Console (Settings → API Keys).

Recommended: generate this connector's API token from a dedicated user with the Read Only role, not from an Account Owner/Admin user. This means that even if the token is ever exposed outside Conduit's control (a credential-store compromise, a misconfigured log, etc.), it cannot be used to delete backups, restores, networks, or devices — matching this connector's own read-only code with a read-only credential, rather than relying on the connector's code as the only safeguard.

Two separate claims, kept at different confidence levels:

• Structurally verified (checked directly, stated with full confidence): this connector's own code makes zero mutating calls (client.ts has exactly one HTTP verb function, doGet), and strips five live secret fields (agent passphrase, IPsec PSK, WireGuard private key, VM VNC password, image-export Samba password) from every response before it reaches a tool result — see secret-redaction.test.ts.
• Vendor-documented, not independently verified (hedged deliberately): whether the Read Only role's restriction is enforced server-side against every mutating endpoint (vs. merely hidden in the console UI) has not been tested against a live Read Only account. This role is also opt-in, not default — a token from an Account Owner/Admin user (the common case for a fresh account) has no such restriction.

Do not read this connector, or this README, as having established that every Slide token is read-only — only that this connector's own code never attempts a mutation, and that Slide offers (but does not default to) a way to make the credential itself read-only as well.

Development

npm installnpm run buildnpm testnpm run lint   # tsc --noEmit

Docker

The image listens on 0.0.0.0 inside the container (MCP_HTTP_HOST) and defaults to AUTH_MODE=gateway. Publish the port on loopback and pass the S2S secret. In gateway mode the Slide token arrives on each request as X-Slide-Api-Token. It is not read from the environment.

docker build -t slide-mcp .docker run \  -e CONDUIT_S2S_SECRET="$CONDUIT_S2S_SECRET" \  -p 127.0.0.1:8080:8080 \  slide-mcp

Compose refuses to start when CONDUIT_S2S_SECRET is unset or empty. The gateway service does not set SLIDE_API_TOKEN and does not use env_file, so a project .env cannot inject a static vendor credential. Compose still interpolates CONDUIT_S2S_SECRET from that file. The published port is 127.0.0.1:8080:8080:

CONDUIT_S2S_SECRET="$CONDUIT_S2S_SECRET" docker compose up

Single-tenant HTTP (AUTH_MODE=env) still requires the S2S secret, then uses SLIDE_API_TOKEN:

docker run \  -e CONDUIT_S2S_SECRET="$CONDUIT_S2S_SECRET" \  -e AUTH_MODE=env \  -e SLIDE_API_TOKEN="$SLIDE_API_TOKEN" \  -p 127.0.0.1:8080:8080 \  slide-mcp

Local HTTP without a secret is development-only. It binds 127.0.0.1 only (a non-loopback MCP_HTTP_HOST, including 0.0.0.0, is ignored and the override is logged) and logs a warning:

MCP_TRANSPORT=http MCP_ALLOW_INSECURE_DEV=1 npm run start:http