MCP-Native

Your IT inventory, operated by your agents.

Flottix is the deterministic registry your AI agents drive through the Model Context Protocol: 17 purpose-built tools, a fully specified REST API, full read/write, with human approval on top. AI proposes. Flottix enforces.

Start Free Pricing
Claude Code
❯ Offboard [email protected] and reclaim all her equipment.

Analyzing [email protected]…

  • • 2 hardware items to reclaim (MacBook Pro, monitor)
  • • 3 license seats to revoke (Slack, Figma, Zoom)
  • • Target status: Pending Inspection

flottix_offboard_employee — awaiting confirmation…

❯

Full access on every plan — Free included

No trial, no credit card, no read-only tier. Sign up, generate a key, connect your agent. We monetize volume and governance — never the protocol.

Setup

Connect Cursor or Claude Code in under two minutes

  1. 01

    Generate an API key

    Settings → API Keys. Granular scopes, revocable anytime.

  2. 02

    Point your client at one URL

    Cursor, Claude Code, VS Code, Cline, Gemini CLI — a remote MCP connection is one URL and one header. No bridge, no extra process.

  3. 03

    Talk to your inventory

    "Which MacBooks are overdue for return?" — your agent answers with real data.

Works with: Cursor · Claude Code · VS Code · Cline · Gemini CLI · any MCP client

Cursor — mcp.json 
{
  "flottix": {
    "url": "https://flottix.app/api/mcp",
    "headers": {
      "Authorization": "Bearer flottix_YOUR_KEY"
    }
  }
}
Claude Code — terminal 
claude mcp add --transport http flottix https://flottix.app/api/mcp \
  --header "Authorization: Bearer flottix_YOUR_KEY"
Programmatic client? Prefer short-lived credentials

Exchange your API key for an OAuth 2.1 token — one hour, audience-bound, scope-narrowed. Discovery at /.well-known/oauth-protected-resourceand instant revocation: delete the key, minted tokens die.

OAuth 2.1 · client_credentials 
curl -X POST https://flottix.app/api/v1/oauth/token \
  -d grant_type=client_credentials \
  -u "flottix_YOUR_KEY:" \
  -d scope="read:assets write:assets"
# → {"access_token": "...", "token_type": "Bearer", "expires_in": 3600}

Claude Desktop or another stdio-only client? The Flottix bridge bridges it — see the agent quickstart.

The Catalog

17 tools. 70+ operations. Zero context bloat.

One typed, verb-dispatched tool per domain instead of dozens of overlapping endpoints — a curated surface that keeps your agent's context lean. Anything beyond it is one flottix_api call away.

Assets

flottix_assets

  • list
  • get
  • create
  • checkout
  • check-in
  • bulk update
  • lifecycle
  • labels

Licenses

flottix_licenses

  • seats
  • assign
  • reclaim
  • renewals
  • compliance
  • bulk update

People

flottix_users

  • directory
  • custody
  • profiles
  • onboard_employee
  • offboard_employee

Kits & Peripherals

flottix_kits · accessories

  • deploy/return kit
  • accessory stock
  • assignments

Parts & Consumables

components · consumables

  • component stock
  • consumable stock
  • detect_zombie_assets

Directory

flottix_directory

  • models
  • categories
  • statuses
  • locations
  • departments
  • custom fields

Insights

flottix_insights

  • dashboard
  • alerts
  • reports
  • cost analysis
  • import/export
  • ingest_procurement

Governance

flottix_governance

  • audit trail
  • physical audits
  • GDPR exports
  • webhooks
  • proposals

flottix_api — the escape hatch

Anything the domain tools don't cover: call any REST endpoint with your identity and scopes. The whole API is specified by a formally valid OpenAPI 3.1 contract your agent can read at flottix.app/openapi.json — interactive docs at /docs.

Compound Workflows

One command instead of twelve round-trips

Real IT operations are multi-step. Compound tools run the whole sequence atomically — with an impact manifest before any write, and the option to file it for human approval instead.

Role-Based Onboarding

❯ Set up [email protected] in engineering with the "Engineering Laptop Bundle" kit.
AI: Account created. Kit assigned: MacBook Pro 14", Dell monitor, keyboard + mouse. 4 items checked out atomically.

Complete Offboarding

❯ Prepare the offboarding of [email protected].
AI: Impact manifest: 2 hardware items to reclaim, 3 license seats to revoke (€47/mo recovered). Confirm to execute — or file it as a proposal for human approval.

Zombie Asset Detection

❯ Is there waste in my inventory?
AI: Found: 4 devices assigned to deactivated users, 6 Figma seats inactive for 60+ days. Waste: €312/mo.

Procurement Ingestion

❯ We just received 30 new laptops — here are the packing slip rows.
AI: 30 serials validated: 28 created, 2 duplicates flagged. Model "Latitude 5450", status Ready to Deploy.
Security by Design

Agents write. The engine decides.

LLMs are probabilistic; your registry cannot be. Every agent write passes through the same deterministic engine, the same tenant isolation, and the same audit trail as humans.

Granular Scopes

Every API key declares exactly what it can read and write — per resource, per action. Revoke anytime.

Two-Phase Confirmation

Sensitive writes first return an impact manifest. The agent must confirm before execution.

OAuth 2.1 Tokens

Short-lived credentials: 1-hour, audience-bound, scope-narrowed tokens exchanged against your key. RFC 9728 discovery built in; revoke the key and minted tokens die instantly.

Immutable Audit Trail

Every action — human or agent — is logged with timestamp, IP, and diff. Exportable on Professional.

Open by construction: MCP · OAuth 2.1 (RFC 9728, 8414, 8707) · OpenAPI 3.1 · llms.txt — no proprietary lock-in.

Give your agents a registry they can operate

Full MCP read/write on every plan. Deterministic state, immutable audit trail, human approval on top.

Start Free See the Agent Tools

Free plan includes the full MCP server. No credit card required.

Immutable Audit Trail
GDPR Compliant
Two-Phase Agent Confirmation