context-viz
por furukama
Web & Frontend DevelopmentResumo
Visualize the current context window usage — token estimates per component (system prompt, tools, workspace files.
Descrição
O que esta skill faz
Estimates and displays a breakdown of the current context window, showing how many tokens each component (system prompt, workspace files, messages, free space) is consuming. Uses a Python script for file token estimation and the session_status tool for live context data. Also lists memory files stored on disk that aren't currently loaded.
Gives a per-component token breakdown rather than just a single usage percentage, so you can pinpoint exactly what to trim.
Quando usá-la
- Checking how full the context is before a long task
- Diagnosing which workspace files are eating the most tokens
- Deciding when to compact or clear context
- Auditing memory accumulation across sessions
- Comparing token usage across different model context sizes
Exemplo de fluxo de trabalho
User asks: how full is my context right now?
- 1
Call session_status to get model, tokens used, tokens total, and compaction count
- 2
Run scripts/estimate_tokens.py against the workspace path to get per-file token estimates
- 3
Subtract file tokens and system overhead from total context used to estimate message tokens
- 4
Calculate free space as total minus used
- 5
Render the monospace bar chart with proportional ░ blocks per component
A formatted context breakdown showing ~Xk tokens per component with bar lengths scaled to percentage of total context, followed by a memory-on-disk inventory grouped by category
Instalação
clawhub install furukama/context-vizExecute o comando da CLI ou cole o prompt no Claude Code, Cursor, Codex ou no agente de código que você usa.