Qortex — AI service layer for agents, REST, and MCP

The service layer

Keep your product focused on the work, not the plumbing.

Qortex puts a durable application boundary around AI calls. Your app or agent selects the capability it needs; Qortex authenticates, routes, meters, and returns a consistent response.

01 / YOUR PRODUCT

Web, mobile, agent, or CLI

Connect Qortex to an application or compatible AI assistant through REST or MCP. Send prompts, message history, tools, or multi-step requests from the runtime you already operate.

your-app → authenticated request

02 / QORTEX

Route, equip, verify

Qortex selects the requested tier, applies account and wallet gates, exposes the right tools, and keeps the work inspectable.

auth → tools → memory → result

03 / YOUR EXPERIENCE

A result your user can use

Return generated content, a verified agent outcome, tool data, or an orchestrated answer without exposing provider details.

result + usage metadata

01 REST APIGenerated docs, authenticated routes, JSON responses, and streaming where supported. 02 MCP serverStreamable HTTP, authenticated tools, and tools/list discovery for capable clients.

Built for useful agents

More than a chat completion.

Use the individual surfaces when you need control, or let the agent harness carry a task from intent to verified result.

01 / AGENTS

Plan, act, and verify

Give an agent a goal and tools, then follow its plan, actions, checks, and result. Supported model choices and fallback rules give your application one consistent way to request AI work.

See the API surface

02 / TOOLS

A built-in tool surface

Add focused functions for code review, documentation, comments, structured articles, and search-informed answers. Combine them with the built-in reading, writing, editing, search, memory, and workflow tools.

Connect through MCP

03 / DISCOVERY

Discover what is live

Discover live capabilities through MCP tools/list and generated REST docs. Keep shared prompts, tool descriptions, and workflow definitions together so applications can reuse the same playbook.

Read the setup guide

04 / ORCHESTRATION

Long-horizon work in stages

Request organized multi-step work for larger tasks instead of coordinating each stage yourself. Qortex carries planning, execution, and checking through an inspectable process.

Explore live docs

05 / MEMORY

Context that carries forward

Carry important earlier and recent messages forward while condensing the middle of a long conversation. Keep a focused history inside the agent or application you already use.

Test in the dashboard

06 / VISIBILITY

Usage you can inspect

Qredits meter AI work across tiers. Check rates, wallet state, and usage views so a useful integration stays understandable as it grows.

View usage

Capability tiers

Name the job. Keep the model out of your code.

Choose the tier that fits the request today. Your integration keeps the same contract while Qortex manages the underlying routing.

01

free

Prototype and explore with zero-cost access.

02

fast

Prioritize low latency for responsive features.

03

economical

Keep high-volume work efficient on cost.

04

adaptive

Use a balanced choice when constraints are ordinary.

05

planner

Decompose complex work before execution.

06

repair

Correct and recover an existing result.

07

max-int

Give complex reasoning more capability and context.

08

orchestrated

Run the gate, plan, parallel-build, and repair stages as one public pipeline.

Q

Qredits are visible by design. Qortex bills AI usage in Qredits (1 qredit = $1), with per-tier rates available in the live docs and usage view.

See rates and usage

One service, two protocols

Choose the interface that fits your runtime.

Call the REST API directly from a product backend, or let an MCP-capable agent connect to the same authenticated capabilities. Both paths use the same service boundary and metering rules.

REST routes and generated reference docs
MCP Streamable HTTP with runtime discovery
Credentials checked before protected tools run
Live details in docs instead of fixed marketing counts

REST documentation MCP setup guide

REST / routeTier

POST https://qortex.forgewise.tech
/api/v1/qortex/routeTier

x-api-key: <your-api-key>
content-type: application/json

{
  "messages": [
    { "role": "user",
      "content": "Review this diff." }
  ],
  "opts": { "tier": "max-int" }
}

MCP / tools

endpoint  https://qortex.forgewise.tech/mcp
auth       x-api-key or Bearer

→ tools/list
← live Qortex tool catalog

→ tools/call
← authenticated tool result

Identity and metering

Authentication that fits the caller.

Keep credentials at the boundary that owns them. Qortex supports Forgewise identity for first-party flows, durable API keys for services, and a documented remote-client path for MCP.

A / SSO

Forgewise SSO

First-party applications can send a Forgewise access token. Qortex validates it through the Forgewise identity surface and can provision the local account on first use.

Authorization: Bearer <access-token>

B / API KEYS

Service credentials

Generate a Qortex API key for direct REST calls or the MCP key for a connected client. Durable app keys keep spend attributable to the app that uses them.

x-api-key: <your-api-key>Get an API key

C / USAGE

Spend you can follow

AI calls pass through the wallet gate and tier pricing. Inspect balance, per-tier rates, and usage history instead of guessing what an integration costs.

Qredits → usage → dashboardOpen usage view

Start in the browser

From first request to a connected agent.

Use the dashboard to test a request, then take the same capability into your application or MCP client.

01

Create a key

Sign in, open API Keys, and create the credential for the app or MCP client that will call Qortex.

Open API Keys

02

Try a tier

Open the Dashboard to test a prompt or tool call. Select a capability tier without wiring a provider into your product.

Test in Dashboard

03

Connect your client

Copy the REST request or MCP configuration from the live docs, then let your app or agent use the same authenticated service.

Configure MCP

04

Watch usage

Review wallet and request activity as you iterate. The usage view keeps the cost of each capability visible.

Review usage

Use cases

A shared foundation for different kinds of work.

↳

Product features

Add generation, review, search, and structured AI responses to a web or mobile product without distributing provider credentials.

⌘

Autonomous agents

Give an agent a goal, tool access, memory, and a verification loop so it can move beyond one-shot answers.

⌁

Coding workflows

Connect a coding CLI or builder agent to authenticated tools and long-running orchestration while your local runtime stays in control.

◎

Research and operations

Use semantic recall, live tool discovery, and staged work for internal research, support flows, and repeatable operations.

Questions, answered

A clear boundary for your AI stack.

When route details or tool schemas change, the live docs and MCP catalog are the source of truth.

Is Qortex a REST API or an MCP server?

Both. Use authenticated REST endpoints when your application owns the request flow, or connect an MCP-capable client to the Qortex MCP endpoint and discover the live tool catalog at runtime.

Do I have to choose a provider or model?

No. Send a capability tier such as free, fast, economical, adaptive, planner, repair, max-int, or orchestrated. Qortex keeps provider and model selection behind that stable designation. The retired default and power designations are rejected rather than remapped.

Are planner and repair the same as orchestration stages?

No. planner and repair are public route tiers you can request directly. orchestrated is the public pipeline tier whose internal gate, plan, build, and repair stages are selected by Qortex.

How do authentication and billing work?

Protected requests accept Forgewise SSO bearer tokens, an x-api-key API key, or a Qortex app key where supported. AI usage is metered in Qredits, and the Dashboard and Usage pages show the resulting spend.

Can an MCP client see tools without a credential?

No. Qortex authenticates the MCP request before exposing or running tools. Get a key in the API Keys page or use the documented Forgewise SSO flow for a remote client.

Ready when you are

Put a real service boundary around your AI.

Start with the dashboard, read the live contract, or connect an MCP client. Qortex gives the next agent or app a place to call.

Open Qortex Set up MCP

Build agents that plan and execute.