What do the different token budget presets do?

Presets like Economy, Standard, and Detailed determine how much context is injected into the AI's prompt, allowing you to balance speed and cost against information depth.

Where is the project memory stored?

All context data is stored locally within a hidden '.claude/memory/' directory inside your project's root folder.

What is the primary purpose of memory-init?

It initializes the ConKeeper system, which provides a structured way for AI agents to store and retrieve project-specific information across different sessions.

Can I share the AI memory with my development team?

Yes. During setup, the skill asks if you want to track the memory files in Git, allowing team members to share the same AI-accessible context.

Does memory-init overwrite existing project documentation?

No, it creates its own dedicated context files designed specifically for AI consumption, complementing your existing READMEs and technical docs.

ConKeeper Memory Initializer

Name: ConKeeper Memory Initializer
Author: swannysec

byswannysec

0•

生产力与工作流

Initializes a persistent, file-based memory system to maintain project context and decision history for AI agents.

The memory-init skill sets up the ConKeeper framework, a specialized directory structure designed to serve as an external 'brain' for Claude within your codebase. By creating structured Markdown files for product context, active tasks, architectural patterns, and glossaries, it ensures that AI agents maintain continuity across development sessions. This skill guides you through project discovery, token budget configuration, and Git integration, transforming Claude from a session-based assistant into a context-aware collaborator that understands your project's history and goals.

主要功能

010 GitHub stars

02Pre-configured templates for tracking decisions, progress, and coding patterns

03Automated creation of structured .claude/memory/ directory and sub-folders

04Interactive project discovery for capturing high-level architecture and goals

05Customizable token budget presets to optimize context loading performance

06Flexible Git integration options for local-only or team-shared project memory

使用场景

01Onboarding Claude to complex legacy codebases with specific architectural constraints

02Ensuring AI context persistence across multiple coding sessions and project restarts

03Maintaining a shared technical decision log that the AI can reference for future tasks

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add swannysec/context-keeper memory-init

For use in Claude.ai and ChatGPT

Download Skill