Aller au contenu
TakoAPI
Retour aux skills

context-compactor

par emberdesire

CLI Utilities
1.1kTéléchargements rapportés par clawskills.sh
0Vues
0 mention J'aimeConnecte-toi pour aimer ce skill
—Pas encore noté
Ta note

Résumé

Token-based context compaction for local models (MLX, llama.cpp, Ollama) that don't report context limits.

Description

Ce que fait ce skill

Client-side token estimation plugin for OpenClaw that proactively summarizes older messages before local models silently truncate or corrupt context. Works with MLX, llama.cpp, and Ollama, which don't report context overflow errors the way cloud APIs do. Configurable thresholds let you tune behavior per model size.

Built-in OpenClaw compaction only triggers when the model reports an overflow error, which local models rarely do, so this plugin catches the problem before it silently corrupts the session.

Quand l'utiliser

  • Running long coding sessions with an 8K MLX model without losing conversation history
  • Debugging why a local model started returning garbled responses mid-session
  • Checking token usage before hitting a context limit with /context-stats
  • Forcing a fresh summary cache when context feels stale
  • Tuning compaction thresholds for a 4K vs 32K local model

Exemple de workflow

ENTRÉE

User sends a message in a long session using an MLX 8K model

AGENT
  1. 1

    before_agent_start hook fires and estimates total context tokens

  2. 2

    Plugin counts ~6,800 tokens against the 8,000 maxTokens threshold

  3. 3

    Splits messages into old (to summarize) and recent (to keep within keepRecentTokens)

  4. 4

    Calls the session model to generate a summary of old messages and caches it

  5. 5

    Injects the summary plus recent messages as the new context before the agent responds

SORTIE

Agent responds normally with full context awareness, without the model ever seeing a truncated or overflowed input

Installation

clawhub install emberdesire/context-compactor

Exécute la commande CLI, ou colle l'instruction dans Claude Code, Cursor, Codex ou l'agent de code que tu utilises.

Liens