Every section of the sidebar,
explained.
Tokalator Pro is a real-time context budget dashboard that lives in your VS Code sidebar. Here is every panel, every metric, and every command it ships with.
Always know where you stand
The sidebar shows your live token count against the model's context window. Budget Level badges (low, medium, high) update in real-time as you open files and accumulate conversation history.
- Tri-state Budget Level: low (<50%), medium (50-80%), high (>80%)
- Tracks open files, pinned files, chat turns, and project size
- Auto-detects active tokenizer (Claude BPE or OpenAI o200k_base)
Look ahead before you send
See exactly how many tokens your next message will add, the new utilization percentage, and how many turns remain before hitting the context ceiling. Prevents surprise context overflow.
- Projects total after next turn: input + estimated output
- Shows utilization delta (e.g. 21.7% of window)
- Estimates remaining turns based on average growth rate
Dollar cost per message
Every turn shows a precise cost estimate based on input tokens, max output tokens, and the active model's per-MTok pricing. Input and output costs are broken down separately so you can see where the money goes.
- Separate input vs output pricing (e.g. $5/MTok in, $25/MTok out)
- 6 model profiles with verified GitHub Copilot billing rates
- Updates live as context grows across the conversation
Track your allowance in real-time
See exactly how much of your GitHub Copilot Pro/Business credits you've consumed. Remaining balance, projected monthly spend, and overage amount are calculated from your actual usage patterns.
- Tracks total used, remaining allowance, and overage
- Monthly projection based on your current usage velocity
- Visual progress bar turns red when credits are exhausted
See what fills your context
The stacked bar decomposes your token budget into five components: open files, system prompt, instructions, conversation history, and output reservation. Instantly identify what's consuming the most space.
- Five components: files, system, instructions, conversation, output reserve
- Color-coded stacked bar for visual proportion
- Conversation history typically dominates after 10+ turns
Find the cheapest model for your context
A live comparison table shows the estimated cost across all 6 model profiles for your current token count. The "vs current" column shows savings or additional cost relative to your active model.
- 6 models: Claude Opus 4.7, 4.6, Sonnet 4.6, GPT-5.5, 5.4, Gemini 3.1 Pro
- Green = cheaper than current, red = more expensive
- Bundled pricing catalog refreshes automatically
Watch your context expand per turn
A per-turn growth chart tracks how your input tokens accumulate across the conversation. Shows the starting size, current size, and average tokens added per turn. Warns when context may be stale.
- Per-turn bar chart from T1 to current turn
- Average growth rate (e.g. +810 tokens/turn)
- Staleness warning after prolonged sessions
Context engineering tools
Pro unlocks four workspace-level tools: security scanning to protect secrets from AI agents, terminology compression to save ~74% overhead, SKILL.md generation for reusable agent expertise, and AI settings generation for GitHub + VS Code configs.
- Secure Workspace: generates .claude/settings.json deny rules for 20 secret patterns
- Generate Glossary: auto-compresses project terminology into .instructions.md
- Generate SKILL.md: packages domain expertise as first-class agent artifacts
- Generate AI Settings: emits .vscode and .github config files
@tokalator Chat Commands
All commands are available through the @tokalator participant in GitHub Copilot Chat.
/countCount tokens in the active file with real BPE/optimizeClose low-relevance tabs to free context budget/pinPin a file to always-include (R = 1.0)/unpinReturn a file to normal relevance scoring/breakdownShow the 5-component token budget breakdown/modelShow or switch the active AI model profile/previewPreview token cost of next message before sending/compactionShow per-turn growth and compaction recommendations/creditsGitHub AI Credits usage, remaining, and projected cost/secureScan workspace security and generate agent settings (Pro)/resetReset session state and chat turn counter