Keeping AI costs down by keeping LLM's 'AIMED' at the target!
Visual Interface for the RAG used in each project
AIMED (AI Monitoring & Execution Dashboard) is a self-contained web interface built on a bundled, modified version of ConPort MCP that provides visual access to project context knowledge used by AI assistants.
The Difference: Read-Only vs. Read-Write Memory Other tools treat AI memory as a read-only cache. AIMED treats it as a read-write document that YOU control — a transparent “Project Brain” you can inspect and correct.
AIMED = ConPort MCP (modified & bundled) + Web Dashboard
Activity Feed - Track recent decisions, progress, and context changes |
Kanban Board - Drag-and-drop task management |
Interactive Knowledge Graph - Visualize project relationships |
Edit Relationships - Right-click to modify connections |
Context Editor - Manage product and active contexts |
Entry Management - View and edit all project entries |
Documentation Hub - Centralized documentation access |
Auto-Generated Docs - View/edit project documentation files |
Workspace Info - Compact view |
Workspace Info - Expanded details |
Progress Tracking - Monitor task completion and status |
|
- How does your AI view your project? Provides insight into what the AI actually sees
- Reduce AI token costs by eliminating context baggage and managing what your AI uses
- Glass-box “Project Brain” — inspect and correct the context your assistant runs on
- Visual knowledge graph showing relationships between decisions, progress, patterns and custom data
- Multi-workspace isolation Ability to work on multiple workspaces at the same time with workspace isolation.
See AIMED-BENEFITS.md for complete features and capabilities.
AIMED uses a modified version of ConPort MCP that enables seamless web interface integration but results in incompatibility with the original ConPort MCP server.
If you're currently using ConPort MCP: Please temporarily disable/turn off your existing ConPort MCP server in your IDE before running AIMED to avoid conflicts.
Custom Instructions Required: Conport and AIMED requires IDE-specific custom instructions for optimal LLM integration. Use the appropriate strategy file from conport-custom-instructions/ directory:
roo_code_conport_strategy- For Roo Code extensioncline_conport_strategy- For Cline extensioncascade_conport_strategy- For Windsurf Cascade (full version)cascade_conport_strategy_compact- For Windsurf Cascade (compact version)generic_conport_strategy- For other MCP-compatible environmentscurrent_generic_conport_strategy- Trimmed down version of generic_conport_strategy that still works well :)mem4sprint.md- Memory and sprint management strategymem4sprint.schema_and_templates.md- Schema and templates for memory management
Copy the entire content of the relevant strategy file into your LLM's custom instructions.
Feel free to use these as a base for your own custom instructions.
→ QUICK_START.md - Condensed copy-paste instructions to get up and running fast
Perfect for experienced users who want minimal explanations and maximum action.
→ INSTALLATION_GUIDE.md - Complete setup instructions with troubleshooting
Includes detailed explanations, troubleshooting steps, advanced configuration options, and platform-specific guidance.
New Project Workflow:
- Turn OFF
conport-aimedMCP server in your IDE - Create project folder + Python environment (venv/conda/etc.) and activate it
- Turn ON
conport-aimedMCP server (createscontext_portal_aimed/automatically) - Add
context_portal_aimed/to.gitignore - Launch dashboard:
python context_portal_aimed/portal_launcher.py
Existing Project:
- Ensure Python environment exists and is activated
- Turn on
conport-aimedMCP server (createscontext_portal_aimed/if missing) - Add
context_portal_aimed/to.gitignore - Launch dashboard:
python context_portal_aimed/portal_launcher.py
Architecture: AIMED uses a modified version of ConPort MCP (equivalent to v0.3.8). It creates context_portal_aimed/ directory for workspace isolation. Each project runs independent server instances with separate port allocation.
- Activity feed showing recent decisions, progress, and context changes
- Kanban board for task management with drag-and-drop functionality
- Interactive knowledge graph displaying project element relationships
- Context editors for product and active project contexts
- Search interface with full-text and semantic search capabilities
- CRUD interfaces for decisions, progress items, system patterns, custom data
- SQLite database with automatic schema migrations
- Model Context Protocol server for AI assistant integration
- Multi-workspace isolation with per-project databases
- Vector embeddings for semantic search
- Full-text search with FTS5
- Knowledge graph with explicit relationship tracking
- Frontend: Next.js 14 with TypeScript, TailwindCSS, TanStack Query
- Backend: FastAPI server with FastMCP integration
- Database: SQLite with Alembic migrations
- Search: FTS5 full-text search + vector embeddings
- Architecture: Multi-workspace isolation with port-based routing
AIMED extends ConPort MCP by GreatScottyMac.
- ConPort MCP: Copyright © 2025 GreatScottyMac and contributors, licensed under Apache License 2.0
- AIMED Visual Extensions: Copyright © 2025 AIMED Contributors
- Current ConPort equivalent: v0.3.8
See NOTICE for complete attribution details.
- Quick Start Guide - Copy-paste instructions to get running fast
- Installation Guide - Complete setup instructions with troubleshooting
- Tips & Tricks - Best practices for working with AI assistants
- Release Notes - Version history
- AIMED Benefits - Complete features and capabilities
- GitHub Issues - Bug reports and feature requests
- AIMED discussion intro - Read what we're about, introduce yourself, share your workflow, and connect with the community
- General discussions - Ask questions, search all topics, start new threads