Stompy – AI Memory For Claude & MCP Agents
Give long-running coding agents durable memory—semantic recall, conflict detection, and lightweight ticketing—so context survives across sessions.
Overview
ai.stompy/stompy is an MCP server for the Build phase that provides persistent agent memory with semantic search, conflict detection, and ticketing.
What is this MCP server?
- Persistent AI memory exposed via hosted MCP at https://mcp.stompy.ai
- Semantic search over stored agent memories
- Conflict detection when new facts contradict prior memory
- Ticketing-style tracking for agent tasks or issues
- Open-source server: github.com/banton/stompy-mcp-server (registry v1.0.0)
- Hosted endpoint: https://mcp.stompy.ai
- Registry version 1.0.0
- Features: semantic search, conflict detection, ticketing (per publisher)
What problem does it solve?
Solo builders re-teach the same project context every session because chat history evaporates and contradictions pile up unnoticed.
Who is it for?
Indie developers running Claude Code or similar agents daily on one codebase who need durable, queryable memory without building a custom vector DB.
Skip if: One-shot scripts, teams forbidden from cloud memory stores, or production incident response that needs full observability stacks instead of agent notes.
What do I get? / Deliverables
After connecting Stompy MCP, your agent can recall, update, and reconcile project memory across sessions with searchable storage and tracked tickets.
- Durable agent-accessible memory store with semantic search
- Conflict-aware updates when project facts change
- Ticket-style tracking for agent-identified follow-ups
Recommended MCP Servers
Journey fit
Agent memory infrastructure is adopted when you harden how your assistant works on the product; Stompy is core agent-tooling in Build, with payoff in Operate as sessions accumulate. Persistent memory, semantic search, and conflict detection are agent-platform concerns, not one-off frontend components.
How it compares
Persistent memory and ticketing MCP, not codebase RAG over embeddings of files or a code-review methodology skill.
Common Questions / FAQ
Who is ai.stompy/stompy for?
Solo builders and agent-heavy developers who want cross-session memory, semantic recall, and light task tracking inside MCP workflows.
When should I use Stompy MCP?
When you adopt long-horizon agent tooling in Build and need memories that persist, searchable context, and detection when facts change.
How do I add ai.stompy/stompy to my agent?
Configure your MCP client with the streamable-http remote https://mcp.stompy.ai per Stompy docs and authenticate as required by the hosted service.