TROVE
A macOS-only local-private runtime that returns bounded cited evidence to external Agents via MCP, with optional Reply Runtime and WeChat source Provider.
README
TROVE
TROVE is a macOS-only, local-private capability runtime. It returns bounded cited evidence to external Agents and can optionally run a local Reply Runtime that generates, reviews, and delivers replies through verified source Providers. Reply delivery is disabled by default. MCP is the primary Agent interface and the CLI is the recovery and operator interface.
The product is TROVE. WeChat support is an optional source Provider rather than the product identity.
Install
From a verified release artifact directory:
python3 -m venv "$HOME/.local/share/trove/runtime"
"$HOME/.local/share/trove/runtime/bin/pip" install ./trove_runtime-1.0.0-py3-none-any.whl ./trove_provider_*.whl
export PATH="$HOME/.local/share/trove/runtime/bin:$PATH"
trove version
Keep the artifact directory and Vault owner-only. Create or select a Vault, then run the redacted health check. The explicit path avoids hidden discovery.
export TROVE_VAULT_ROOT="$HOME/Trove/trove-vault"
mkdir -p "$TROVE_VAULT_ROOT"
chmod 700 "$TROVE_VAULT_ROOT"
trove --vault "$TROVE_VAULT_ROOT" doctor
Connect MCP
Register trove-mcp through Agent Switch with these arguments:
--pack standard --vault $TROVE_VAULT_ROOT
Run agent-switch doctor before changing its central configuration and
agent-switch reconcile afterward. Do not hand-edit generated client configs.
The standard pack is sufficient for ordinary recall and search.
First call
Ask the Agent to call trove_recall, or use the exact CLI fallback:
trove --vault "$TROVE_VAULT_ROOT" recall --target "Example person" --limit 50
The JSON envelope states ok, typed errors, citations, and coverage. Follow an
opaque cursor only when the requested coverage needs another page.
Failure path
Run trove --vault "$TROVE_VAULT_ROOT" doctor. Retry only when
error.retryable is true. For ambiguous_target, select one returned account.
For approval_required, stop: an Agent can request or inspect approval but only
a human at the controlling terminal can decide it.
See MCP, operations, and the generated capability reference. Provider setup is separate; see the installed source Provider. The optional Reply Runtime architecture and safety model are documented in Reply Runtime.
Privacy boundary
Real chat databases, exports, media, transcripts, provider payloads, secrets, logs, and local Vault data do not belong in this repository. Tests and checked-in evidence must be synthetic or explicitly source-safe. Run the privacy scanner before every commit.
See open-source privacy and security policy.
Contributing
See CONTRIBUTING.md. TROVE is licensed under Apache License 2.0.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.