Tempr Chat

One chat panel. Any provider. Your keys, your models.

Tempr is a bring-your-own-key (BYOK) AI coding assistant for Visual Studio 2022 (17.14 or later) and Visual Studio 2026. Add your own API key for a first-party provider (OpenAI, Anthropic, Google, xAI, and 20+ more) or an aggregator like OpenRouter, and pick whichever model fits the task — nothing routed through a shared pool, no rate-limit surprises, no vendor lock-in.

🤖 Agent mode, your way

A real tool-calling loop that reads, searches, and edits your codebase and runs allow-listed commands — with three control levels:

• Interactive — approve every change before it lands
• Bypass — auto-apply edits, you review after
• Autopilot — hands-off multi-step execution

Every file the agent touches lands in the Total Changes panel, with a diff per file and Undo All / Accept All, so you review a whole turn in one place.

The agent is debugger-aware (inspects the live call stack and locals, evaluates expressions mid-session), test-aware (runs your tests with structured pass/fail results), and solution-aware (live compiler diagnostics, project references, call graph).

Prefer a plain conversation? Ask mode answers with your active file, solution overview, and package dependencies as context.

🎭 Personas & multi-agent delegation

Type @debugger, @test, @security, @review, @docs, @refactor, @modernize, @profiler, @git, or @plan for a specialist prompt and tool scope — or define your own custom personas (Pro).

Mid-turn, the agent can hand focused sub-tasks off to specialist personas — e.g. delegating root-cause analysis to @debugger while it keeps working — running several in parallel when it helps.

🔍 Real semantic code search, precise context

Embedding-based search over your solution finds conceptually related code even when the wording doesn't match — not just grep with extra steps.

• #file and #symbol attach exactly the code you mean
• @workspace, @terminal, and @problems fold in a solution overview, terminal output, or the Error List
• Images — attach a screenshot from the + menu or paste one straight into the chat

⚡ Inline completion & rich chat

Ghost-text completions suggest the next line as you type — press Tab to accept.

• Multiple chat tabs, each with its own model, mode, and history
• Send to Tempr — the light bulb on any error or warning sends it to chat with the surrounding code attached
• Slash commands (/explain, /fix, /tests, /doc), one-click copy and insert, regenerate, and searchable chat history

🔑 Bring a key from 30+ providers

Keys are encrypted at rest and scoped to your license — never pooled into a shared quota. Requests are relayed through Tempr to reach your chosen provider, and usage bills straight to your own provider account, so there's no markup on tokens.

🧩 Extensible & synced

• MCP support — browse a curated catalog of Model Context Protocol servers or add your own; servers assigned to your account in Tempr's vault show up automatically
• Cloud sync — sessions, config, custom personas, and keys (encrypted) follow you across machines when you're signed in
• Budget alerts — set an optional monthly budget and get a heads-up as you approach it

📊 See your usage in the Tempr Portal

What's new in 2.2.28

• Reasoning and Thinking level implemented

Getting started

1. Install the extension and restart Visual Studio
2. Start a 7-day free trial at temprhq.io — your license key is emailed instantly
3. Open the chat from View › Tempr (or Ctrl+\, Ctrl+O) and enter your license key in Settings
4. Add your provider API key(s), pick a model, and start chatting, completing, and delegating

Plans start at $15/mo (limited-time introductory pricing). Every plan includes the full model catalog with your own keys. Teams plan coming soon.

Privacy

Unexpected errors are reported automatically so bugs get fixed quickly — error message and stack trace only, never your source code or chat content. Turn it off anytime in Tools › Options › Tempr › General.

Questions? temprhq.io · FAQ · support@temprhq.io