Overview / Description
Clawtex is an AI developer tool that gives Claude AI agents a persistent, structured memory layer so they stop forgetting context between sessions and repeating the same mistakes. It works across all Claude surfaces, including web, desktop, mobile, and CLI, and organizes memory into three layers. State holds queryable structured entities such as clients, projects, and goals that represent what's currently true. Events are timestamped, typed records of decisions, milestones, and actions logged across every session. Lessons are AI-extracted patterns from recurring mistakes, turned into structured guardrails with enforcement steps so the agent corrects course over time. Clawtex exposes native tool calls for logging events, updating state, and querying history, so agents use it without manual API configuration. It is aimed at Claude users running multi-session projects, teams running multiple agents that need shared context, and professionals who want decision tracking with a continuous improvement loop. Because it is built specifically around Claude, it is not a general-purpose memory layer for other LLM providers.
Used For
Giving Claude AI agents persistent, structured memory across multi-session projects and agent workflows
Pricing
Pros & Cons
Pros
- Three-layer memory model: State (entities), Events (timestamped records), Lessons (extracted guardrails)
- Works across all Claude surfaces — web, desktop, mobile, and CLI
- Native tool calls for logging, updating state, and querying history with no manual API setup
- AI-extracted lessons turn recurring mistakes into enforced guardrails
Cons
- Built specifically for Claude; not a general memory layer for other LLMs
- Free tier caps usage at 1 agent, 200 events/month, and 14-day history
- Unlimited event history and AI lesson extraction require a paid plan
Questions & Answers
Alternatives
Mem0, Zep, Letta (MemGPT)