vstack ships an MCP (Model Context Protocol) server that exposes all 34 diagnostic patterns as MCP tools, plus per-pattern citations + playbooks + composition manifests as resources, plus invocation templates as prompts.
Local stdio subprocess. Zero hosting cost on the vstack side. Compatible with any MCP-aware client.
pip install 'valanistack[anthropic,mcp]'Claude Desktop — paste into ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):
{
"mcpServers": {
"vstack": {
"command": "vstack-mcp",
"args": ["serve"],
"env": {"ANTHROPIC_API_KEY": "sk-ant-..."}
}
}
}Cursor — same JSON into ~/.cursor/mcp.json (or project-level .cursor/mcp.json).
Cline / Continue / Roo / Windsurf / Zed — open the MCP-servers config in the extension settings and paste the same block.
Or auto-generate the snippet for any platform:
vstack-mcp config-snippet claude-desktop
vstack-config gen-platform cursor # extends to 14 platforms- 34 tools — one per pattern, named
vstack_<pattern_name>(e.g.vstack_lewin,vstack_aar,vstack_schein_culture). - 102 resources —
vstack://patterns/index(catalogue),vstack://patterns/<name>/citations,vstack://patterns/<name>/playbooks,vstack://patterns/<name>/composition. - 35 prompts — meta
vstack_pick_patternrouter + onevstack_<name>_invoketemplate per pattern.
Once configured, ask your client to run any of the 34 patterns by name — "Use the Schein culture audit on this trace…" — and the client picks the matching tool from the catalogue, the server runs the analyzer, and the detection comes back as structured JSON. The server runs as a local stdio subprocess; nothing leaves your machine except whatever LLM calls the analyzer itself makes.
vstack-mcp serve # the daemon (your MCP client launches this)
vstack-mcp list-tools # all 34 tool names + friendly labels
vstack-mcp list-resources # all 102 resource URIs
vstack-mcp config-snippet <client> # generate the config JSON for a client