HyperStack — Developer-Controlled Knowledge Graph Memory for AI Agents

Your agent controls the graph, not an LLM hallucination.

HyperStack gives AI agents persistent memory with a knowledge graph. Store knowledge as typed cards (~350 tokens) with explicit linked relations. Retrieve via hybrid semantic + keyword search or graph traversal. Time-travel to debug past decisions.

Why HyperStack?

• Explicit control — Your agent defines cards and links directly. No LLM auto-extraction, no phantom relationships, no extraction cost.
• Time-travel debugging — Query the graph at any timestamp. See what your agent knew when it made a bad decision. "Git blame for agent memory."
• Zero LLM cost — Mem0 and Zep charge ~$0.002 per memory operation (LLM extraction). HyperStack: $0.
• 30-second setup — One API key, one env var. No Neo4j, no Docker, no OpenSearch.
• 94% token savings — ~350 tokens per retrieval vs ~6,000 tokens stuffing full context.

Quick Start

1. Sign up at https://cascadeai.dev/hyperstack (free, 50 cards)
2. Set env vars:

export HYPERSTACK_API_KEY=hs_your_key
export HYPERSTACK_WORKSPACE=default

3. Store a card:

curl -X POST "https://hyperstack-cloud.vercel.app/api/cards?workspace=default" \
  -H "X-API-Key: $HYPERSTACK_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"slug":"use-clerk","title":"Use Clerk for Auth","body":"Chose Clerk over Auth0","cardType":"decision","links":[{"target":"alice","relation":"decided"}]}'

Features

• Knowledge graph with 9 typed relations (owns, decided, triggers, blocks, depends-on, etc.)
• Graph traversal — trace decisions, dependencies, ownership across linked cards
• Time-travel — reconstruct the graph at any point in time using version history
• Hybrid search — semantic (pgvector) + keyword matching
• Auto-extract — pattern-based fact extraction from text (no LLM, free)
• Visual graph explorer — interactive canvas with force-directed layout
• Team workspaces — shared memory across multiple agents

Also Available As

• MCP Server: npx hyperstack-mcp
• Python SDK: pip install hyperstack-py
• JavaScript SDK: npm install hyperstack-sdk
• OpenClaw Plugin: npm install openclaw-hyperstack

Pricing

| Plan | Price | Cards | Features | |------|-------|-------|----------| | Free | $0 | 50 | All features — graph, impact, replay, identity | | Pro | $29/mo | 500+ | Branching, webhooks, priority support | | Team | $59/mo | 500 | Everything in Pro + collaboration | | Business | $149/mo | 2,000 | Everything in Team + scale |

Links

• Website: https://cascadeai.dev/hyperstack
• API: https://hyperstack-cloud.vercel.app