HODORSpendesk × Hodor
Spendesk MCP, through Hodor
Spendesk doesn't have an official MCP server. Hodor exposes the Spendesk API as Model Context Protocol tools — with identity, policy, audit, and a kill switch — so finance copilots can safely categorize expenses, approve under threshold, and manage virtual cards.
Spendesk is a European spend management platform combining cards, expense claims, invoice processing, and approval workflows.
The gap
Why Spendesk doesn't have an MCP — yet
Spendesk has a public API, but no official Model Context Protocol server — agents have no safe, standardized way to call it.
Finance data is sensitive. A raw API token gives an agent access to every transaction, virtual card, and budget across the org.
Without per-agent policy, a finance copilot could approve over-threshold expenses, issue cards without limits, or expose vendor data — with no attribution when audit season comes.
How it works
Spendesk as MCP, in three steps
- STEP 01
Connect Spendesk to Hodor
Authenticate once with your Spendesk account. Hodor stores the credential securely and never exposes it to agents.
- STEP 02
Hodor exposes the API as MCP tools
Every Spendesk endpoint becomes a typed MCP tool. Scope per agent, set policy, and define rate limits at the gateway.
- STEP 03
Agents call MCP through Hodor
Any MCP-compatible agent — Claude, Cursor, Dust, n8n, custom — connects to Hodor. Every call is checked, logged, and attributed.
Tool catalog
What your agents can do
A subset of the Spendesk tools Hodor exposes as MCP. Enable, disable, or constrain each one per agent identity.
- spendesk_list_expenses
- spendesk_get_expense
- spendesk_categorize_expense
- spendesk_approve_expense
- spendesk_reject_expense
- spendesk_create_virtual_card
- spendesk_freeze_card
- spendesk_get_budget
- spendesk_list_invoices
- spendesk_export_to_accounting
Built for production
Identity, policy, audit — by default
The same controls Hodor applies to every integration apply to Spendesk: per-agent identity, scoped tools, real-time policy enforcement, full audit logs, and a global kill switch.
Agent identity
Every agent gets a unique, revocable identity. Every call is attributed.
Scoped policy
Fine-grained tool access, rate limits, and field-level restrictions enforced at the gateway.
Full audit trail
Every call logged with payload, identity, and policy outcome — SOC 2 / ISO 27001 ready.
Kill switch
Revoke any agent in one click. Hodor blocks all downstream calls instantly.
Common patterns
What teams build with Spendesk + Hodor
Auto-categorize & flag
Agent reviews new expenses, applies the right category, and flags anything outside policy — read + categorize only, no approval rights.
Manager approval copilot
Approve expenses under a configurable threshold automatically, escalate anything above. Hodor enforces the threshold at the gateway, not in the prompt.
Vendor onboarding agent
Issue scoped virtual cards for new SaaS subscriptions — with a hard monthly limit per card and zero permission to modify existing cards.
Month-end close bot
Read-only agent reconciles Spendesk transactions against accounting and posts a digest of unresolved items. No write access to Spendesk at all.
FAQ
Spendesk MCP, answered
Does Spendesk have an official MCP server?
+
Not as of 2026. Spendesk exposes a REST API and webhooks, but no native Model Context Protocol server. Hodor bridges this by exposing Spendesk endpoints as MCP tools, callable from Claude, Cursor, Dust, n8n, or any custom agent runtime.
Can I cap how much an agent can approve?
+
Yes. Hodor policies let you set a maximum approval amount per agent, per category, or per vendor. The cap is enforced at the gateway — the agent literally cannot call the approve endpoint above the threshold.
Are virtual card creations safe to delegate to an agent?
+
With Hodor, yes. You scope an agent to a specific card-issuance flow with a hard monthly volume and per-card spending limit. Every issued card is attributed to the originating agent identity in the audit log.
What audit evidence do I get?
+
Every Spendesk call made via Hodor is logged with agent identity, full payload, policy outcome, and timestamp. Logs are exportable for SOC 2, ISO 27001, and internal finance reviews.
How do I get started?
+
Book a demo. We'll connect Spendesk, scope the finance tools you want to delegate, and walk through policy thresholds and audit logs live.
Ship your Spendesk agents — safely.
Identity, policy, and audit for every Spendesk call your agents make. Set up in under an hour with a Hodor engineer on the call.
- 20-minute demo
- No credit card
- SOC 2-ready logs