Cortega for Developers

Govern coding assistants and model access.

Give developers effective, lower-cost AI with open-weight models running on-premises or in private hosting. Keep familiar tools like Cursor, Claude Code, and Codex. Start with the Cortega AI Pack for VS Code on Windows or macOS for supported assistants. Add Endpoint Guard on macOS when you need Shadow AI controls.

Lower AI costsYour code stays privateNo manual tool setup
Request a demo
Your tools. Your rules. Your models.
Cursor Claude Code Codex CLI
Configure supported tools with Cortega
Cortega AI Gateway
Team budgetsRoute or lock modelsData guardrails
Requests go to your approved models
Open-weight LLMsOn-premises
Open-weight LLMsPrivately hosted

VS Code extension: Windows & macOS. Endpoint Guard: macOS.

Why Cortega for developers

Coding assistant routing, budgets and policy

01 / COST

Make AI affordable for the whole team.

Use cost-effective open-weight models for coding and debugging. See spend by team and developer, then set budgets and limits for traffic through Cortega.

Control team spend →
02 / PRIVACY

Keep source code in your environment.

Run models on-premises or in dedicated private hosting. Choose where code and prompts are processed, and apply data guardrails to approved traffic.

Explore private deployment →
03 / SIMPLICITY

Set it up once. Roll it out to everyone.

Start with Cortega AI Pack on Windows or macOS to configure your supported coding tools. Cortega applies model access and team budgets to their routed requests.

See supported coding tools →
Cortega AI Pack · Developers, lawyers and other teams

Project attribution and usage costs

Use one matter/project list across your tools. Let the Matter/Project agent assign confident conversation matches, review uncertain suggestions, and see the cost of governed AI usage in Cortega Agents → Insights.

01 / ASSIGNMENT

Connect conversations to matters and projects.

Add project descriptions such as repositories, products and development tasks, or legal matter details such as clients and key facts. The agent rechecks ongoing conversations after a pause, assigns confident matches, and keeps uncertain suggestions for review. Human decisions take priority. You can also choose a matter/project explicitly from the extension.

02 / INSIGHTS

See AI spend by project.

Compare governed LLM request counts, tokens and usage costs by matter/project in Cortega Agents → Insights. Explicit request selections take priority over conversation assignments. Unmatched usage is shown separately, so the report does not hide unassigned costs.

03 / SHARED TOOLS

Manage your project list from your assistant.

Cortega AI Pack automatically registers a shared Matters/Projects MCP server in Claude Code and Codex at sign-in. Use project or matter tools to list, add, update and bulk-import entries, with permissions and audit records. Turn on Privilege/sensitive work when requests must use a Sensitive choice in your team's model authorization.

Automatic assignments apply to captured conversation usage; they do not change your selection for unrelated work. Useful descriptions and conversation history help the agent distinguish projects. The Matter/Project agent must be licensed and enabled for your organization.

You choose how models are used

Route for the task. Or lock to one model.

Apply routing rules to teams, individual developers, and production agents.

Request-based routing

Match each request to a model suited to the work.

Routine coding requestLower-cost model
Complex reasoning requestReasoning model

Illustrative rules: you define the routing policy.

Enforced model locking

Assign a model once. Every request uses it.

Team, developer, or production agentRequests any model
Cortega enforces the assigned model

The caller’s model selection does not override the lock.

Explore model routing and controls →
Endpoint Guard · macOS

Add device controls when you need Shadow AI protection.

On macOS, Endpoint Guard configures supported coding tools and monitors or blocks unapproved AI on enrolled devices. Add it when your policy needs to cover AI use beyond configured gateway connections.

Explore Endpoint Guard →
Choose a policy for each tool
Approved coding toolConfigure & route
Unapproved AI toolMonitor
Restricted AI toolBlock access

Policies apply on enrolled macOS devices.

Choose your setup

Keep the tools your team already knows.

Use Cortega AI Pack for VS Code on Windows and macOS to configure the coding tools below, plus Continue. Add Endpoint Guard on macOS when you need Shadow AI monitoring or blocking.

Start with Cortega AI Pack

Windows & macOS

Configure Claude Code, Codex, Continue, Cursor, OpenCode, Pi, Goose, and Aider from VS Code. Standalone assistants are discovered without needing their VS Code plugins.

Close Cursor before running Sync, then reopen it. Its supported BYOK connection uses saved Cursor settings.

Add Endpoint Guard

macOS

Configure the coding tools below on enrolled Macs. Add device-level monitoring or blocking when you need to govern unapproved AI tools.

The VS Code extension configures supported connections; Endpoint Guard adds Shadow AI controls.

The VS Code extension also supports Continue. Model and API compatibility depend on the tool and model. Cursor’s BYOK requests still pass through Cursor’s servers; configuring private models does not make Cursor fully local.

See Endpoint Guard controls →
People collaborating around laptops at a shared table
Photo by Annie Spratt / Unsplash

AI FOR DEVELOPMENT TEAMS

Configure developer tools and approved models.

Keep familiar coding assistants connected to approved models, with shared access policies, project visibility, and controls over AI costs.

MAKE YOUR NEXT MOVE

Private deployment for coding assistants

Bring open-weight models, team cost controls, and managed coding tools to your developers.

Request a demo Learn more Explore the Cortega documentation