SPECMEM
- web-chat
Persistent semantic memory for Claude Code, chats AND code. 74+ MCP tools, local pgvector embeddings, codebase indexing with 14 language extractors, multi-agent team coordination. Self-hosted, zero cloud dependency.
About
What is SPECMEM?
SpecMem is a persistent memory layer for Claude Code that stores conversations, indexes your codebase, and runs semantic search entirely on your hardware. It is an MCP server designed for developers who want their coding assistant to remember project context across sessions without any cloud dependency.
How to use SPECMEM?
Install SpecMem globally via npm with sudo npm i -g specmem-hardwicksoftware, then configure it as an MCP server for Claude Code. Once connected, your assistant automatically retains knowledge of your architecture, bugs, and decisions from previous sessions.
Key features of SPECMEM
- 74+ MCP tools for memory and codebase interaction
- 14 language extractors for deep code understanding
- Indexes 30,000 files in under 3 minutes
- Zero API costs – runs completely locally
- Persistent session memory across coding sessions
- Free for individual developers earning under $35k/year
Use cases of SPECMEM
- Fix a bug without re‑explaining your authentication setup
- Continue a feature from a previous Claude Code session seamlessly
- Search your entire codebase semantically without manual query tuning
- Maintain architectural knowledge without maintaining a CLAUDE.md file
FAQ from SPECMEM
What makes SPECMEM different from other MCP servers?
SpecMem provides 74+ MCP tools, while the closest competitor offers only 4. It also includes 14 language extractors and can index 30,000 files in under three minutes.
Does SpecMem send any data to the cloud?
No. SpecMem runs entirely on your hardware with no API costs. Nothing goes to the cloud.
What platforms or editors does SPECMEM support?
SpecMem is built for Claude Code and runs locally on your machine via an npm installation.
Is SPECMEM free to use?
It is free for individual developers making under $35,000 per year. It is built by Hardwick Software Services.
Are there any known limits?
The README does not mention limits beyond the free tier income threshold. The tool indexes 30,000 files in under 3 minutes and uses local hardware resources.
Details
- Author
- jonhardwick-spec
- Category
- web-chat
- Repository
- jonhardwick-spec/specmem
Every AI coding session starts from scratch. your assistant forgets your architecture, your bugs, your decisions. i got tired of re-explaining my codebase so i built something about it.
SpecMem is a persistent memory layer for Claude Code. it's an MCP server that stores your conversations, indexes your codebase, and runs semantic search locally. nothing goes to the cloud.
what that looks like in practice: you say "fix the auth bug" and claude already knows your auth setup because it remembers from last session. no re-explaining, no CLAUDE.md maintenance.
some numbers:
- 74+ MCP tools (closest competitor has 4)
- 14 language extractors
- 30,000 files indexed in under 3 min
- zero API costs, runs entirely on your hardware
free for individual devs making under $35k/yr. built by Hardwick Software Services.
sudo npm i -g specmem-hardwicksoftware
