Awareness – Cloud Memory for AI Agents | Multi-Agent Shared Memory via MCP

@EdwinHao-EverestAn

a month ago

# memory ai-agent mcp claude cursor claudecode windsurf cline github-copilot persistent-memory multi-agent workflow-rules vector-search semantic-search knowledge-graph developer-tools rag context-management

One-line MCP config gives your AI agent persistent memory with automatic workflow rules. Unlike passive memory stores, Awareness teaches your agent WHEN to remember and recall — works across Claude Code, Cursor, Windsurf, Cline, Copilot and 7 more IDEs.

Overview Tools Comments

Overview

Why Awareness, not just another memory server?

Most memory MCP servers are passive stores — they save and retrieve text. Awareness is an active memory system with three unique capabilities:

1. Auto-injects workflow rules into your IDE

Rules are written to your IDE config (CLAUDE.md, .cursorrules, etc.) so your agent automatically inits memory at session start, recalls before decisions, and records after changes. No manual prompting needed.

2. Structured knowledge extraction — not flat text

Every memory is auto-classified into Knowledge Cards, Action Items, and Risks with conflict detection (new / update / duplicate / contradiction).

3. Multi-agent shared memory

Multiple agents (Claude Code + Cursor + Manus) share one memory library. Agent A's decisions are instantly available to Agent B.

Quick Start

Sign up at awareness.market and get your API key
Add the MCP config above to your client
Or even faster: npx @awareness-sdk/setup (auto-configures MCP + workflow rules for 12 IDEs)

MCP Tools

Tool	Description
`awareness_init`	Load full session context: 7-day narratives, knowledge cards, open tasks, active skills, agent profiles, synthesized rules — all within your token budget
`awareness_record`	Save memories with inline structured insights (knowledge cards + action items + risks). Auto-deduplicates and detects conflicts with existing knowledge
`awareness_recall`	Hybrid search: vector semantic + keyword full-text + fusion + diversity reranking. 5 modes (precise/session/structured/hybrid/auto), configurable token budget
`awareness_lookup`	Fast structured queries (<50ms): tasks, knowledge cards, risks, session history, timeline, handoff context, dependency graph, rules, agent profiles
`awareness_get_agent_prompt`	Spawn sub-agents with role-specific activation prompts and memory context — enables multi-agent workflows

Awareness vs other memory servers

Feature	Generic Memory MCP	Mem0	Awareness
Persistent storage	✅	✅	✅
Semantic search	✅	✅	✅ Hybrid (vector + full-text + reranking)
Auto workflow rules	❌	❌	✅ Auto-injected into IDE
Structured insights	❌	❌	✅ Knowledge cards + risks + action items
Conflict detection	❌	❌	✅ new/update/duplicate/contradiction
Multi-agent sharing	❌	Limited	✅ Role-based agent profiles
Token budgeting	❌	❌	✅ Smart context trimming
IDE support	1-2	2-3	12 IDEs
Setup	Manual	SDK required	`npx` one-line CLI

Key Features

Persistent memory across AI sessions and tools
Hybrid retrieval — vector + full-text search for accurate recall
Workflow rules — auto-injected into IDE, agents use memory without prompting
Insight extraction — automatically surfaces decisions, risks, and action items
Conflict detection — flags contradictions across memory entries
Multi-agent collaboration — shared memory with role-based access
SDK support — Python and TypeScript SDKs for custom integrations

Supported AI Tools

Claude Code · Cursor · Windsurf · Cline · GitHub Copilot · Codex · Kiro · Trae · Zed · JetBrains · Augment · Gemini CLI · Manus · ChatGPT · Devin

Pricing

Plan	Price	API Calls	Storage
Free	$0/mo	1,000/mo	500 MB
Starter	$5/mo	10,000/mo	5 GB
Pro	$12/mo	100,000/mo	50 GB
Team	$29/mo	500,000/mo	200 GB
Enterprise	Custom	Unlimited	Custom

Server Config

{
  "mcpServers": {
    "awareness": {
      "url": "https://awareness.market/mcp",
      "headers": {
        "Authorization": "Bearer <YOUR_API_KEY>"
      }
    }
  }
}

Build with ShipAny.