Vardoger
Personalize how your coding agent behaves by analyzing local conversation history on your machine, without sending that history to a cloud trainer.
Overview
vardoger is a Build-phase MCP server for agent-tooling that personalizes coding assistants by analyzing local conversation history on your machine.
What is this MCP server?
- Runs locally—analyzes conversation history on your machine for personalization signals
- PyPI package v0.3.1 with stdio MCP via vardoger mcp entrypoint
- VARDOGER_MCP_PLATFORM override for cursor, claude-code, codex, copilot, windsurf, cline, openclaw
- Multi-host support so one install can align with whichever agent you use this week
- PyPI package version 0.3.1
- Supported platform overrides: cursor, claude-code, codex, copilot, windsurf, cline, openclaw
- Transport: stdio with positional mcp argument
Community signal: 3 GitHub stars.
What problem does it solve?
Every new agent session starts cold even though your machine already holds months of chats that encode how you like code written and reviewed.
Who is it for?
Privacy-conscious solo builders who live in multiple AI coding hosts and want persistent, local personalization across sessions.
Skip if: Teams needing centralized shared memory in the cloud, or developers who never store local chat logs on disk.
What do I get? / Deliverables
After installing vardoger over stdio MCP, assistants can draw on locally derived personalization instead of generic defaults alone.
- MCP-accessible personalization derived from local history
- Platform-aware behavior via optional VARDOGER_MCP_PLATFORM
- On-machine processing without requiring a hosted memory SaaS
Recommended MCP Servers
Journey fit
Primary shelf is agent-tooling because the product improves the agent loop itself; personalization pays off again in ship review and operate iterate. Agent-tooling is where MCP servers that read host chat logs and adapt assistant behavior belong in the solo-builder catalog.
How it compares
Local chat-memory personalization MCP, not a cloud team wiki or a single-repo CLAUDE.md generator.
Common Questions / FAQ
Who is Vardoger for?
It is for individual developers using Cursor, Claude Code, Codex, Windsurf, or similar tools who want assistants adapted from their own local conversation history.
When should I use Vardoger?
Use it when starting build work, reviews, or iteration and you want the agent to reflect prior local sessions instead of behaving like a first-time collaborator.
How do I add Vardoger to my agent?
Install Vardoger from PyPI (v0.3.1), configure stdio MCP with the mcp subcommand, optionally set VARDOGER_MCP_PLATFORM to your host, then reconnect your client.