prism-coder
by dcostenco·★ 133·Score 49
MCP server providing persistent memory, semantic search, and local LLM for AI agents.
Overview
Prism Coder is a comprehensive Model Context Protocol server that creates a 'Mind Palace' for AI agents, enabling persistent memory across sessions. It implements cognitive routing with different memory stores (episodic, semantic, procedural) and supports both cloud and local inference through its own fine-tuned models. The system features semantic search capabilities, proactive session drift detection, and a visual dashboard. Prism operates locally-first with HIPAA-hardened architecture, requiring no API keys for the free tier, and supports both Ollama local models and cloud fallback options.
Try asking AI
After installing, here are 5 things you can ask your AI assistant:
When to choose this
Choose Prism Coder when you need AI agents with persistent memory across sessions and require offline functionality without API dependencies.
When NOT to choose this
Avoid if you need cloud-based AI processing, require real-time collaboration features, or need extensive write capabilities beyond its memory system.
Tools this server exposes
9 tools extracted from the READMEsession_load_contextRecover prior session's state on boot
session_save_ledgerAppend immutable session log entry
session_save_handoffSave live state for the next session
knowledge_searchSemantic + keyword search over all memories
query_memory_naturalNatural-language Q&A over your Mind Palace
extract_entitiesPull people / projects / decisions from text
session_synthesize_edgesAuto-link related memories into a graph
session_cognitive_routeCompare current work against original goals
session_compact_ledgerCompress and reload only what matters when drifted
Comparable tools
Installation
Install Prism Coder using npm:
# Install globally
npm install -g prism-mcp-server
# Or use npx (no install)
npx prism-mcp-serverAdd to Claude Desktop / Cursor config:
{
"mcpServers": {
"prism": {
"command": "npx",
"args": ["-y", "prism-mcp-server"]
}
}
}Download local models with Ollama:
ollama pull dcostenco/prism-coder:1b7 # 2.2 GB
ollama pull dcostenco/prism-coder:8b # 4.7 GB
ollama pull dcostenco/prism-coder:14b # 8.4 GB
ollama pull dcostenco/prism-coder:32b # 19 GBFAQ
- How does Prism Coder handle data privacy?
- Prism operates locally-first with HIPAA-hardened architecture. The free tier runs entirely on your machine with SQLite and local embedding models, requiring no API keys or cloud connections. Data never leaves your device unless explicitly synced to the cloud tier.
- Can I use Prism Coder offline?
- Yes, the free tier runs entirely offline with local models. Prism supports an offline cascade system (14B → 8B → 1.7B) that maintains functionality even without internet access, ensuring critical tools like AAC (Augmentative and Alternative Communication) remain available.
Compare prism-coder with
Last updated · Auto-generated from public README + GitHub signals.