AiTA: AI Token Analyzer
Publisher: Chapar.AI
Website: https://chapar.ai
AiTA is a product of Chapar.AI.
Local-first VS Code extension for analyzing AI agent token usage. AiTA scans the chat history that Codex, Claude Code, Gemini CLI, and GitHub Copilot store on your machine, and shows where tokens go by provider, user, model, repository, task type, and skill. All data stays local — nothing is uploaded.
Requirements
Install the extension from the Visual Studio Marketplace. No repository clone is required.
AiTA bundles the analyzer runtime and installs its runtime dependencies into VS Code global storage on first use. Node.js 20.19+ and npm must be available on your PATH so the local SQLite dependency can be installed for your OS/CPU.
Features
- Activity Bar view — click the AiTA icon to see a summary plus quick actions (open dashboard, scan, report). The dashboard opens automatically the first time the view is shown.
- Dashboard in VS Code or browser — provider tabs (All / Codex / Claude / Copilot / Gemini / Other), user scope, date ranges, model trends, and skill impact. The local server is started on demand and reused if already running.
- Scan — parses all four agent histories into a local SQLite database.
- Reports and exports — markdown report in the editor, CSV/markdown export, model drift / pricing match table.
Commands
AiTA: Scan AI Agent History — scan Codex, Claude Code, Gemini CLI, and Copilot history, then show a report snapshot.AiTA: Show AI Token Analyzer Report — open the current report as markdown.AiTA: Open Dashboard in VS Code / AiTA: Open Dashboard in BrowserAiTA: Show Model Drift Report — model names, normalized keys, and pricing match status.AiTA: Export Data — CSV or markdown export via the AiTA terminal.AiTA: Show Effective Config — resolved settings as JSON.AiTA: Select User Scope — switch between consolidated and per-user reporting.AiTA: Refresh Summary — refresh the sidebar view.
Settings
| Setting |
Default |
Description |
codexTokenAnalyzer.repoRoot |
"" |
Optional analyzer repository checkout for extension development. Empty = use bundled runtime. |
codexTokenAnalyzer.dbPath |
"" |
SQLite database path. Empty = .token-analyzer/token-analyzer.sqlite under the active runtime folder. |
codexTokenAnalyzer.codexDir |
"" |
Codex home directory. Empty = ~/.codex. |
codexTokenAnalyzer.claudeDir |
"" |
Claude Code home directory. Empty = ~/.claude. |
codexTokenAnalyzer.geminiDir |
"" |
Gemini CLI home directory. Empty = ~/.gemini. |
codexTokenAnalyzer.copilotDir |
"" |
VS Code workspaceStorage directory for Copilot chats. Empty = platform default. |
codexTokenAnalyzer.dashboardPort |
4317 |
Local dashboard server port. |
codexTokenAnalyzer.defaultRange |
1m |
Default report/dashboard date range (1d/1w/1m/ytd/1y/max). |
codexTokenAnalyzer.defaultProvider |
all |
Default provider filter. |
codexTokenAnalyzer.defaultScope |
all |
Default scope (all or user). |
codexTokenAnalyzer.defaultUserKey |
"" |
Default user key when scope is user. |
codexTokenAnalyzer.nodeExecutable |
node |
Node executable used to run the analyzer CLI. |
Privacy
Scanning happens locally and results are written to a local SQLite file. Raw transcript text is inspected in memory for signals (skills, tools, outcomes) but is not persisted. Secrets, emails, and tokens are redacted before any optional snippet storage.
The packaged VSIX is written under packages/vscode-extension/.
