Where is the distilled information stored?

The skill writes results to a local journal and updates specific entity files grouped by topic, creating a structured knowledge base within your project directory.

Does it include tool calls and system messages in the summary?

No, the skill specifically filters out tool calls, results, and system messages to focus only on the actual human-assistant dialogue and technical outcomes.

How does it handle very large conversation files?

Distill acts as a coordinator that spawns smaller sub-agents to process individual files, ensuring that the main context window doesn't get overloaded with raw transcript data.

What does the Distill skill do for Claude Code?

It scans your recent conversation history to find important facts, decisions, and completed tasks, then saves them to a persistent journal so the AI remembers them in future sessions.

Memory Distiller

Name: Memory Distiller
Author: ascorbic

byascorbic

•

108

•

生産性とワークフロー

Extracts actionable knowledge and facts from daily AI conversations to maintain persistent, self-refining project memory.

The Distill skill provides a sophisticated memory consolidation system for Claude Code by scanning local conversation logs and extracting critical technical decisions, project-specific facts, and completed actions. By utilizing a coordinator-sub-agent architecture, it efficiently processes large volumes of interaction data without overwhelming the primary context window. This ensures that important project knowledge, specific developer preferences, and complex technical trade-offs are preserved in a long-term knowledge base rather than being lost when conversation history is cleared.

主な機能

01Automated journaling of completed actions and technical outcomes

02108 GitHub stars

03Smart filtering of system messages and tool calls for content relevance

04Scalable sub-agent architecture to prevent context window blowup

05Automated fact extraction from daily .jsonl conversation logs

06Persistent entity file updates for long-term project knowledge

ユースケース

01Building a project-specific knowledge base that evolves automatically during development

02Consolidating developer preferences and style guides for consistent AI behavior

03Tracking technical architectural decisions and their rationales across long cycles

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add ascorbic/macrodata distill

For use in Claude.ai and ChatGPT

Download Skill