AgentIndex icon
AgentIndex
ToolsMCP ServersHOTCategoriesTrendingNewBest ListsCompareGuides
Submit Tool
ToolsMCP ServersHOTCategoriesTrendingNewBest ListsCompareGuides
Home/
LLM Infra/
conduit
conduit logo

conduit

Active·★ 233·MIT·Updated 2026-10-25
★ Hidden Gem

Conduit acts as a local Model Context Protocol (MCP) gateway, designed to drastically reduce token overhead when AI agents utilize multiple tools by consolidating server tools into a few meta-tools. It simplifies the setup and authentication of various services, allowing them to be configured once and used across all connected AI clients.

conduit sits in the LLM Infra category, so it is usually better judged by workflow fit, deployment style, and team needs than by isolated features alone. Based on the available metadata, it leans most clearly toward ~90% fewer tokens by using lazy-discovery mode for tools.. A representative use case is Reducing token costs and improving context window efficiency for AI agents using many tools.. The listed license is MIT. GitHub stars are currently 235. Relevant tags include AI Agent, Tool Management.

This write-up is based on the project's README and public repository data. We have not hand-tested the tool.

#AI Agent#Tool Management#Token Optimization#Context Management#API Gateway#LLM Infrastructure#Workflow Optimization
$ Install
$ git clone https://github.com/tsouth89/conduit.git && cd conduit && npm install && npm run build:gateway && npm run tauri dev
↗ Visit site★ GitHub
01

Features

01~90% fewer tokens by using lazy-discovery mode for tools.
02Single setup and authentication for all AI clients.
03Per-agent scoping to control tool access for different clients.
04Built-in governance to toggle tools on/off and audit calls.
05A built-in playground to test tools before integration.
02

Why choose it

+~90% fewer tokens by using lazy-discovery mode for tools.
+Single setup and authentication for all AI clients.
+Reducing token costs and improving context window efficiency for AI agents using many tools.
+Supports 5 platforms or runtime environments.
03

Trade-offs

!The final decision should still be made through side-by-side comparison with similar tools.
04

Compatibility

Windows
OS
Verified via docs
macOS
OS
Verified via docs
Linux
OS
Verified via docs
Model Context Protocol (MCP)
Protocol
Verified via docs
AI Clients
Integration
Verified via docs
05

Quick start

1
$ git clone https://github.com/tsouth89/conduit.git
2
$ cd conduit
3
$ npm install
4
$ npm run build:gateway
5
$ npm run tauri dev
06

Use cases

↳Reducing token costs and improving context window efficiency for AI agents using many tools.
↳Managing and authenticating multiple external services (e.g., Stripe, Supabase, GitHub) for various AI clients from a single gateway.
↳Providing granular control over which tools different AI clients or agents can access, enhancing security and compliance.
↳Testing and validating AI tools (e.g., from an MCP server) using a built-in playground before deploying them to agents.
07

Alternatives

⚡Top VS:
VSmcp-compress-router★21→VSharnessrouter★3.7k→VSllmtrim★257→VScodex-pooler★233→
mcp-compress-router logo
mcp-compress-router★ 21
Compress all connected MCP into a single router MCP and save up to 99% on tokens
VSCompare→
harnessrouter logo
harnessrouter★ 3.7k
HarnessRouter Community Edition: the self-hosted, Apache-2.0 edition of the unified interface for agent harnesses. Run Codex, Claude Code, Hermes, PI, DSH, and more through one API, with sessions, streaming, files, cancellation, and failure handling. Implements the Unified Harness Protocol (UHP), an open standard. Your keys, your infrastructure.
VSCompare→
llmtrim logo
llmtrim★ 257
Local proxy that compresses your LLM API requests so you pay less, with no change to the answers. Trims wasted tokens from prompts, history, tool output, and code before they're sent: -31% input / -74% output, measured live. Any provider, no extra model calls. Also an MCP server and embeddable library (Rust, Python, Ruby, Kotlin, Swift).
VSCompare→
codex-pooler logo
codex-pooler★ 233
🤖 Pool and route Codex accounts behind one Gateway
VSCompare→
toolport logo
toolport★ 233
Local-first MCP gateway. One port for every tool and every AI client: lazy discovery (~90% token savings), tool integrity + quarantine, secrets in the OS keychain.
VSCompare→
tachyon logo
tachyon★ 37
💫 Tachyon MCP Runtime for Java and Kotlin - Model Context Protocol (MCP) server with Streamable HTTP, native Netty transports, tools, tasks, resources, prompts, completions, resumable sessions, and stateless deployment support.
VSCompare→
projectmem logo
projectmem★ 870
Local-first AI coding memory for AI agents. Records issues, attempts, fixes and decisions, then warns your agent before it repeats an approach that already failed. Native MCP server for Claude Code, Cursor, Antigravity and Codex. 100% local, no cloud, no telemetry. MIT.
VSCompare→
clarilayer logo
clarilayer★ 76
Stop re-explaining your data to your AI every session. The individual-analyst context layer, delivered over MCP (Claude Code / Cursor / Codex).
VSCompare→
See all alternatives →

Comments

Log in to leave a comment

No comments yet. Be the first!

On this page
01Features02Why choose it03Trade-offs04Compatibility05Quick start06Use cases07Alternatives
tsouth89 /
conduit
↗

Local MCP gateway that cuts your agent's tool-token overhead ~90%. Every MCP server's tools collapse to 3 meta-tools the agent searches on demand. Measured: 97% less per request, same results. Keys in your OS keychain, no cloud.

233MIT
Last commit3d ago
StatusActive
CategoryLLM Infra
Links
Documentation↗Discussion↗Issues↗Releases↗
Cloud Infrastructure
Ad

Deploy on DigitalOcean

Host your MCP servers, Docker containers, and AI agents with $200 free credit over 60 days.

Claim $200 Free Credit
© 2026 AgentIndex.app|Built by a 10-year iOS Developer.
Best by use caseAboutContactAdvertisePrivacy PolicyTerms of UseQYSGitHubBuy me a coffee ☕

Browse by Category

MCP ServersCode AssistantWorkflow AutomationRAG / Knowledge BaseMulti-AgentBrowser AutomationLLM InfraDev ToolingObservability

Not affiliated with Anthropic, OpenAI or Microsoft.