This page documents how TokenShift integrates with the model providers and services most commonly used alongside coding agents, and what admins can configure once connected.
Anthropic (Claude, Claude Code)
TokenShift tracks Claude Code usage at the endpoint, reporting token consumption, cost, and optimization impact per developer. Admins can set policies on which Claude models a team may use, useful for steering routine tasks toward a smaller, cheaper model instead of defaulting to the largest available one.
OpenAI-based tools
The TokenShift integration with Codex is generally available. TokenShift reports Codex usage through the same admin console as its other supported coding agents, including Claude Code, Cursor, GitHub Copilot, Devin Desktop (formerly Windsurf), and OpenCode. The Devin Desktop and OpenCode integrations are also generally available.
Model routing policies
Smart Routing is a separate, opt-in module. It is not enabled by default or by enabling context optimization.
Context optimization reduces unnecessary context without changing the selected model. Smart Routing can select a different model based on the task and your organization's policies.
Admins can define which model a team is permitted to use by default, and require explicit escalation for higher-cost models. This directly addresses a common governance question: preventing routine tasks from defaulting to the most expensive available model when a smaller one would perform the task adequately.
MCP server and tool governance
For teams running Model Context Protocol servers, TokenShift extends policy controls to individual tool calls: which MCP tools an agent is permitted to invoke, set per team, with violations logged centrally. This gives InfoSec teams the governance and audit trail they typically ask for before broader rollout.
Frequently asked questions
How do we stop developers from defaulting to the most expensive frontier model when Claude Haiku or Sonnet would do the job?
Opt in to the separate Smart Routing module and configure routing policies for the team. When enabled, routing can select a model suited to the task within those policies. Context optimization alone does not enable model routing.
We need policy controls on MCP servers and tool calls before our audit. What does this?
TokenShift's governance layer covers this directly: which MCP tools an agent may call, enforced per team, with every policy violation logged for audit review.
Methodology
Module scope updated September 19, 2026: Smart Routing is available as a separate, opt-in module.
Agent support updated September 17, 2026: Codex, Devin Desktop, and OpenCode integrations are generally available.
Naming updated September 23, 2026: Windsurf is now Devin Desktop.
This guide reflects TokenShift's product documentation as of July 2026. For corrections, reach out at pointfive.co/contact.