Recallium Concepts

MiniMe: Community version—free forever, private, self-hosted.

How Recallium Works

Recallium is a persistent memory system for AI coding assistants. Unlike traditional AI tools that forget everything between sessions, Recallium automatically captures your development context as you work, organizing it into searchable memories that persist forever. MiniMe is the Community version built for Developers.

The Developer Journey

When you use Recallium MiniMe with your AI coding assistant (Claude Desktop, Cursor, VS Code, Windsurf, or JetBrains), the AI intelligently stores important context as you develop. Here's how it works:

1

You Code & Chat with AI

Work normally with your AI assistant. Implement features, fix bugs, make architectural decisions, have technical discussions. No special commands needed.

2

AI Automatically Stores Context

Your AI assistant intelligently captures important moments: code implementations, decisions made, bugs fixed, insights discovered. The AI knows what's worth remembering and stores it automatically.

3

Context Persists Across Sessions

Days, weeks, or months later, when you return to the project, your AI remembers everything. Past decisions, implementation details, and tribal knowledge are instantly available.

4

Search & Recall Instantly

Ask your AI questions like "Why did we choose PostgreSQL?" or "How did we implement authentication?" and get instant answers from your project's memory.

5

Discover Patterns & Insights

MiniMe automatically clusters related memories and generates insights across your entire development history. Discover recurring patterns, identify technical debt, track learning progression, and surface best practices without manually reviewing every memory.

6

Build Knowledge Graphs

Your memories form interconnected knowledge graphs that connect related concepts, decisions, and implementations across projects. Visualize relationships, trace dependencies, and understand how different pieces of your codebase relate to each other.

Memory Types

As your AI assistant works with you, Recallium organizes information into 11 different memory types. Each type serves a specific purpose, making it easy to find exactly what you need later. Your AI automatically chooses the right type based on context—you don't need to think about it.

code-snippet

What it stores: Implementation details, code changes, technical decisions about specific files.

Example: "Implemented JWT authentication with refresh token rotation using Redis for storage. Access tokens expire in 15 minutes, refresh tokens in 7 days."

Always linked to specific files so you can find past implementation notes before editing code.

decision

What it stores: Architectural choices, technology selections, and the reasoning behind them.

Example: "Chose PostgreSQL over MongoDB because we need ACID transactions for financial data and complex relational queries for reporting."

Answers "why did we..." questions instantly, preventing repeated debates.

rule

What it stores: Coding standards, preferences, constraints, and team conventions.

Example: "Always use TypeScript strict mode. Prefer functional components over class components. All API calls must include retry logic."

Acts as guardrails for your AI, ensuring consistent code quality across all suggestions.

learning

What it stores: Insights, "aha" moments, lessons learned, and discoveries.

Example: "Discovered that useEffect dependency arrays must include all external variables to prevent stale closures. Missing dependencies cause subtle bugs."

Builds your team's collective knowledge, preventing the same mistakes twice.

debug

What it stores: Bug investigations, root cause analysis, and troubleshooting notes.

Example: "Login fails with 500 error. Stack trace shows null pointer in session validation. Root cause: missing Redis connection check. Fixed by adding connection validation before session lookup."

Helps recognize and fix recurring issues faster by surfacing past solutions.

design

What it stores: Architecture explorations, system design discussions, and technical proposals.

Example: "Exploring caching strategies for the API: Redis for session data (fast, TTL support), Memcached for API responses (distributed, simple), considering tradeoffs between consistency and performance."

Captures the design thinking process, not just the final decision.

research

What it stores: Investigation findings, external references, and gathered information.

Example: "Research on OAuth2 PKCE flow: RFC 7636 defines Proof Key for Code Exchange to prevent authorization code interception. Recommended for mobile and SPA applications."

discussion

What it stores: Team conversations, meetings, code reviews, and feedback.

Example: "Team discussion on microservices migration: Decided to start with a modular monolith first. Concerns about operational complexity and team size. Will revisit in 6 months."

progress

What it stores: Status updates, milestones, and checkpoint summaries.

Example: "Week 3 sprint: Completed authentication module (login, signup, password reset). Starting API integration next week. Blocked on database schema approval."

task

What it stores: Simple action item notes and TODOs.

Example: "TODO: Add error handling to login endpoint. Need to handle network failures and invalid credentials gracefully."

For formal task tracking with status management, use the dedicated Tasks feature instead.

working-notes

What it stores: General notes, thoughts, and temporary information (default type).

Example: "Investigating performance issue in dashboard component. Load time is 3+ seconds. Suspects: large data fetches, unnecessary re-renders, missing memoization."

The default catch-all type when other categories don't fit.

Projects & Documentation

Beyond individual memories, Recallium helps you organize formal project documentation. While memories capture day-to-day work, Projects store high-level deliverables like product requirements, architecture plans, and project briefs.

What are Projects?

Projects are containers for formal documentation that stakeholders and team members reference. Think of them as your project's "source of truth" documents that evolve over time.

Project Briefs

What it stores: High-level overview documents describing project goals, scope, and current status.

Your project's executive summary that stakeholders reference for quick understanding.

Product Requirements (PRDs)

What it stores: Detailed specifications of what needs to be built, including features and acceptance criteria.

Formal documentation that defines what the system will do and how success is measured.

Implementation Plans

What it stores: Technical architecture and development roadmaps showing how the system will be built.

Technical blueprints that guide the development team through the implementation process.

Project Relationships

What it stores: Links between related projects to understand dependencies and discover related work across your codebase.

Connect projects together to see how different systems and initiatives relate to each other.

Tasks

What it stores: Actionable work items with status tracking, priorities, and completion notes linked to project documentation.

Formal task management with lifecycle tracking, linking related memories and documentation for complete context.

Document Uploads

What it stores: Uploaded documents (PDFs, specs, manuals) with tech-specific chunking and encoding for intelligent search and retrieval.

Documents are automatically processed with semantic chunking and vector encoding, making them searchable alongside your memories.

Your AI can create and update these documents automatically by synthesizing information from your working memories, or you can explicitly ask it to document the current state of your project.

Insights & Pattern Analysis

Recallium doesn't just store memories—it analyzes them to discover patterns, trends, and reusable approaches across your entire development history. This is where the real intelligence emerges.

What are Insights?

Insights are AI-powered meta-analyses that examine all your memories to find recurring patterns, identify technical debt, track learning progression, and surface best practices. Your AI runs these analyses automatically and can answer questions like:

"What patterns emerged in our API error handling?"

Recallium analyzes all your error-handling code across projects and discovers: "Common pattern: Centralized error middleware in 3 projects. Custom error classes (ValidationError, AuthError) used consistently. Circuit breaker pattern emerging in recent services."

"Why do we keep having authentication bugs?"

Identifies recurring root causes: "Race conditions from concurrent token refresh (3 occurrences). Fix: Request queuing works consistently. Token expiry edge cases (2 occurrences). Fix: Add 2-minute buffer for refresh trigger."

"Show me reusable approaches for caching"

Surfaces proven patterns: "Redis for session data (5 implementations), Memcached for API responses (3 implementations), LRU cache for computed values (2 implementations). Success rate: 90% reduction in database load."

Types of Insights

Pattern Analysis

What it discovers: Recurring approaches, design patterns, and architectural decisions across all your work.

Surfaces reusable solutions and consistent approaches that have proven effective across projects.

Quality Analysis

What it discovers: Recurring bugs, root causes, and successful fix approaches to prevent future issues.

Identifies patterns in bugs and their solutions, helping you avoid repeating the same mistakes.

Technical Debt

What it discovers: Code smells, incomplete implementations, and areas needing refactoring based on your notes.

Highlights accumulated technical debt and suggests prioritization for refactoring efforts.

Learning Progression

What it discovers: How your understanding evolves over time, showing knowledge growth and skill development.

Tracks your learning journey, showing how your approaches and solutions improve over time.

Progressive Disclosure

Recallium uses an intelligence-first approach that starts with summaries and expands to details only when needed. This keeps your AI fast and focused while still giving access to comprehensive context.

1. Search

What it does: Find relevant memories using natural language queries. Results show summaries and key information—enough to decide what's relevant.

Start with high-level overviews before diving into details.

2. Expand

What it does: Get full details only for the specific memories you need. This saves time and keeps AI responses focused.

Drill down into complete context only when necessary.

3. Analyze

What it does: Ask for insights to see patterns across all memories without reading each one individually.

Get meta-analysis and pattern recognition across your entire knowledge base.

4. Refine

What it does: Update or enhance memories as your understanding grows, keeping your knowledge base current.

Maintain an evolving knowledge system that improves over time.