How does Sage automatically recall knowledge?

Sage uses keyword-based triggers defined in an index. When your query matches these triggers, the relevant knowledge items are automatically injected into your context window.

Can I store knowledge specific to a single project?

Yes, Sage supports project-scoped registries located in a .sage/knowledge directory within your project, in addition to global storage in your home directory.

How do I handle outdated information?

Sage provides specific tools to update, deprecate, or archive knowledge items, ensuring your AI doesn't rely on obsolete architectural decisions or logic.

What is the best way to save new insights?

You can save knowledge by using the sage_save_knowledge MCP tool, the Sage CLI, or simply by telling Claude to 'remember this' after a significant discovery.

Is there a limit to how much knowledge can be recalled?

While multiple items can be recalled at once, Sage manages this additively up to a reasonable token limit to ensure context remains useful without being overwhelmed.

Sage Knowledge Management

Name: Sage Knowledge Management
Author: b17z

byb17z

•

Productivity & Workflow

Persists and recalls research insights, architecture decisions, and domain expertise across AI chat sessions using semantic checkpoints.

Sage Knowledge is a sophisticated context preservation skill for Claude Code that enables the creation of a persistent knowledge base for your AI workflows. It allows developers to 'checkpoint' AI research, store validated technical decisions, and recall critical project information automatically through natural language triggers. By bridging the gap between disconnected sessions, it ensures that Claude maintains a long-term memory of technical constraints, user preferences, and complex logic, effectively eliminating context loss and the need for repetitive prompting during complex development cycles.

Key Features

01Automatic context injection based on natural language keyword triggers

02Transparent reporting showing which items were recalled and their token impact

031 GitHub stars

04Comprehensive lifecycle management including deprecation, archiving, and updates

05Multi-interface access via MCP tools, CLI, and natural language conversation

06Dual-scoped storage for both global and project-specific knowledge bases

Use Cases

01Preserving complex architectural decisions and the 'why' behind technical choices

02Facilitating seamless session handoffs by saving the current state and prior research

03Storing domain-specific regulatory summaries or API constraints for recurring reference

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add b17z/sage sage-knowledge

For use in Claude.ai and ChatGPT

Download Skill