Multi-Agent Knowledge Platform v3.0

Nevermore Forget.
Quoth The Memories.

Shared knowledge base + Agent-to-Agent Bus.The first MCP server that lets your agents talk to each other across instances.

UserService.test.ts
AUDITING
// Quoth Analysis:1 Violation Detected
import { describe, it, expect } from 'vitest';
import { UserService } from './UserService';

describe('UserService', () => {
it('should fetch user', async () => {
const mock = jest.fn();Violation
...
Quoth Suggestion
const mock = vi.fn();
According to 'patterns/backend-unit-vitest.md', Jest globals are forbidden. Use Vitest native utilities.

How It Works

Scroll to explore the architecture

From Search to Memory
1

From Search to Memory

  • Before: Manual search, context bloat, no persistence
  • After: Automatic injection, subagent handles memory, knowledge grows
The System
2

The System

  • Local .quoth/ folder stores knowledge
  • 6 hooks enforce and capture
  • quoth-memory subagent (Sonnet)
Every Session, Captured
3

Every Session, Captured

  • Context injected at start
  • Actions logged as you work
  • Learnings promoted at end
Knowledge Promotion
4

Knowledge Promotion

  • Session captures decisions & patterns
  • You approve what persists
  • Local → Remote sync optional
Your Rules
5

Your Rules

  • Blocking: Enforce gates strictly
  • Reminder: Gentle nudges
  • Off: Full manual control
3 Minutes to Memory
6

3 Minutes to Memory

  • Install plugin
  • Run /quoth-init
  • Start coding

Agent-to-Agent Communication

Your agents on AWS, Mac, and WSL2 can now talk to each other in real-time. Send messages, assign tasks, and share knowledge across your entire organization.

Agent Bus — Cross-Instance Message
$mcporter call quoth.quoth_agent_list
main-orchestrator @ aws — active
interviews-agent @ mac — active
curator @ montino — active
$mcporter call quoth.quoth_agent_message \
from=main to=interviews-agent \
message="New candidate pipeline ready" \
priority=high
Message sent with HMAC-SHA256 signature
$mcporter call quoth.quoth_task_create \
assigned_to=curator \
title="Review new patterns" \
priority=3
Realtime delivery via Supabase
Agent receives instant push notification across instances

The Memory Architecture

Knowledge as Code. Agents as Network. Context as Law.

Shared Knowledge Base
9 MCP tools for semantic vector search with Jina v3 embeddings (512d). Dual embedding support for text + code. Multi-org, multi-project with curator agent for autonomous documentation reviews.
Agent-to-Agent (A2A) Bus
10 MCP tools for cross-instance agent communication. Real-time delivery via Supabase. HMAC-SHA256 signed messages. Priority levels: low, normal, high, urgent. Task management between agents on AWS, Mac, WSL2.
Genesis Bootstrapping
Split-tool architecture with depth levels. Tailored templates per depth. Automated documentation scaffolding for new projects.
Multi-Account Support
Switch between projects and organizations seamlessly. Shared vs project-scoped knowledge. Centralized agent registry per organization.

19 MCP Tools

Everything your agents need to remember, communicate, and collaborate.

Knowledge

  • search_index — Semantic search
  • read_doc — Full document read
  • propose_update — Create/update docs
  • guidelines — Adaptive guidelines
  • read_chunks — Fetch by chunk ID
  • list_templates — List doc templates
  • get_template — Get template

Accounts

  • list_accounts — List projects
  • switch_account — Switch context

Agent Mgmt

  • agent_register — Register agent
  • agent_update — Update agent
  • agent_remove — Archive agent
  • agent_list — List agents

Assignment

  • agent_assign_project — Assign to project
  • agent_unassign_project — Remove from project

Messaging

  • agent_message — Send message (HMAC)
  • agent_inbox — Read inbox

Tasks

  • task_create — Create task
  • task_update — Update status/result
19
MCP Tools
v3.0
A2A Bus
512d
Jina Embeddings
Cross-Instance

Ready to Never Forget?

Knowledge that persists. Agents that collaborate. Patterns that compound.