Ryan Cuff
Ryan Cuff is an embedded-systems and hardware/integration engineer who builds software and hardware projects in public. His main current project, Woolly, is local, private project memory for coding agents: facts cite the code they came from and are flagged when that code changes.
At a Glance
- Individual software developers
- Software engineering teams
- Users of Claude Code, Cursor, Codex, Gemini CLI, VS Code, Windsurf, Claude Desktop, LM Studio and opencode
- Developers building AI agents and apps
- +1 more
AI Tools by Ryan Cuff
(1)Woolly
Local Memory for Coding Agents
Discussions
No discussions yet
Be the first to start a discussion about Ryan Cuff
Latest News
Woolly 0.15.0: code-grounded project facts and stale-fact detection became the central release capability.
Woolly 0.14.1 improved semantic search with embedding-model prefixes and reported updated LOCOMO retrieval results.
Woolly 0.14.0 launched a mem0-shaped local memory API, memory history/versioning and dreaming.
Woolly 0.13.0 launched project memory in git via woolly init, with shared .woolly Markdown memory and agent configuration export.
Products & Services
A local-first memory system and CLI for coding agents. It stores readable Markdown memories, links project facts to files/functions/line ranges and commits, checks those references after code changes, flags stale facts, shares project memory through git, and connects to multiple AI apps through MCP or a proxy.
A local API shaped like mem0's routes and field names. It can infer facts from conversations, add/search memories, preserve history when facts change, and scope memories by user, agent, run or project; JavaScript and Python packages are available.
A local proxy that adds memory to Ollama and apps using the OpenAI or Anthropic APIs.
Market Position
Woolly positions itself against generic conversation-memory tools by tying each project fact to source code and flagging it when the code changes. Its main differentiation is local-first, readable Markdown memory shared across agents and teammates, versus hosted memory APIs; the README explicitly references mem0 as the API shape and says hosted memory APIs are a better fit for products with thousands of users.
Leadership
Founders
Ryan Cuff
Embedded systems engineer and hardware/integration engineer. His public profiles describe work spanning hardware integration, QA/QC, UI/UX and software development; his personal site documents building hardware from scratch and writing embedded C. He is the creator and maintainer of Woolly on GitHub under rjcuff.
Executive Team
Ryan Cuff
Creator and maintainer
Embedded systems engineer; public GitHub profile identifies him as rjcuff and links Woolly and ryancuff.com.
Founding Story
Woolly was started to solve a recurring problem Ryan Cuff describes from switching among Claude Code, Cursor and Codex: each agent starts with no project context, while memory tools can preserve facts that the code no longer supports. The initial vision was a local, shared, readable memory whose facts cite source code and become stale when that code changes, without an account, cloud service or telemetry.
Business Model
Revenue Model
Woolly is distributed as a free, MIT-licensed, local open-source project. The public materials describe no hosted service, account requirement or paid subscription; the project is monetized neither through cloud/API usage nor disclosed enterprise tiers.
Pricing Tiers
Local open-source software under the MIT license; no account, cloud or telemetry is required.
Target Markets
- Individual software developers
- Software engineering teams
- Users of Claude Code, Cursor, Codex, Gemini CLI, VS Code, Windsurf, Claude Desktop, LM Studio and opencode
- Developers building AI agents and apps
- Privacy- and local-first AI users
- Developers switching among multiple AI coding agents on the same repository
- Teams that want agents to share project decisions, conventions, gotchas and commands through git
- Privacy-sensitive developers who want local memory without a hosted service
- Builders adding memory to their own agents or applications through the API and SDKs
- Ollama, OpenAI-API and Anthropic-API users who want a local memory proxy