skills.rest
javicasper
javicasperCOMMUNITY
Explore all skills by this author →
GitHub

memory-forge

Extract and route actionable knowledge from work sessions into CLAUDE.md or AGENTS.md files.

Updated Jan 28, 2026
Search Tags:#routing#monorepo#semantic-search#knowledge-extraction#autoload#cli-agnostic
One-click install
npx skills add https://github.com/javicasper/memory-forge --skill memory-forge
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: memory-forge
Source: https://github.com/javicasper/memory-forge/tree/main
Command: npx skills add https://github.com/javicasper/memory-forge --skill memory-forge

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Memory Forge automates extraction and organization of actionable knowledge from work sessions, helping AI agents capture non-obvious learnings and route them into the right long-term memory locations while keeping autoload context lightweight.

Core Features & Use Cases

  • Monorepo aware: routes knowledge to the appropriate module or root context (CLAUDE.md / AGENTS.md) or to the knowledge directory.
  • CLI-agnostic: works with Claude Code, OpenCode, Codex, Cursor, and any Agent Skills-compatible tool.
  • Autoload vs indexing: supports autoloaded skill/context files and indexed knowledge for semantic search when MCP is present.
  • Knowledge lifecycle: extract, organize, and preserve patterns, fixes, and conventions from debugging sessions for reuse.

Quick Start

Install Memory Forge by placing the SKILL.md in your agent's supported skill location and activate the skill after completing a task to begin extracting knowledge.

Frequently Asked Questions about memory-forge

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I capture actionable knowledge from AI agent work sessions for reuse?▼

Automating knowledge extraction from work sessions captures non-obvious learnings and routes them into appropriate long-term memory locations like CLAUDE.md or AGENTS.md files. This preserves debugging patterns and conventions for future reuse while keeping autoload context lightweight.

How does knowledge routing work in a monorepo with multiple AI agents?▼

Monorepo knowledge routing directs extracted learnings to the correct root CLAUDE.md, module-specific CLAUDE.md, AGENTS.md, or a dedicated knowledge directory. This ensures context is loaded only where relevant, optimizing the agent's working context across different project modules.

Can I use agent skills for knowledge extraction across different CLI tools?▼

Yes, CLI-agnostic agent skills provide extraction and routing guidance that works across Claude Code, OpenCode, Codex, and Cursor. The skill operates independently of the specific CLI, ensuring knowledge capture functions consistently regardless of the agent tool used.

What is the difference between autoloaded context and indexed knowledge for AI agents?▼

Autoloaded context involves files like CLAUDE.md automatically loaded into the agent's prompt, keeping immediate context lightweight. Indexed knowledge uses semantic search via MCP to retrieve specific information on demand, preventing context bloat while preserving deep project knowledge.

Do I need MCP to organize and retrieve extracted agent knowledge?▼

MCP is not required to organize extracted knowledge, as the skill routes learnings to standard context files like CLAUDE.md. However, MCP is necessary to enable semantic search indexing, which optimizes context loading by querying a knowledge directory rather than autoloading everything.

Related Skills

Explore More Software Engineering Skills→
LusciiLuscii

make-skill-template

Generate SKILL.md frontmatter and starter folder structure for new Agent Skills.

Official
Basic
bryonjacobbryonjacob

documenting-with-claude-md

Organizes and loads CLAUDE.md documentation hierarchies for AI context.

Community
Basic
dingdyandingdyan

openviking

Serve RAG and semantic search over documents via an OpenViking MCP interface.

Community
Intermediate
AlejandroFiginiAlejandroFigini

learn

Capture and organize project learnings into a per-project JSONL file.

Community
Intermediate