Agentcents runs on your own machine — the client starts it, so there is no endpoint to ping. 196 installs a week from pypi. Last commit 25 Jul 2026.
Local LLM cost proxy: forecasts each call and recommends the cheapest model that clears the task.
Nothing serious here today
Today is the operative word: we check Agentcents every 15 minutes and re-read its code on every release. Watch it and you find out the day that stops being true.
Connect this server
This server runs on your own machine — install it with the package manager and the client starts it for you. Package name taken from the official registry entry.
claude mcp add agentcents -- uvx agentcents{
"mcpServers": {
"agentcents": {
"command": "uvx",
"args": [
"agentcents"
]
}
}
}[mcp_servers.agentcents]
command = "uvx"
args = ["agentcents"]{
"mcpServers": {
"agentcents": {
"command": "uvx",
"args": [
"agentcents"
]
}
}
}{
"servers": {
"agentcents": {
"type": "stdio",
"command": "uvx",
"args": [
"agentcents"
]
}
}
}Alternatives to Agentcents
same job, measured the same wayCut LLM token costs: count tokens, estimate cost, slim prompts, and pick the cheapest capable model.
Offline MCP: predict LLM call cost and pick the cheapest capable model. No keys, no network.
Zero-cost MCP server for local code inspection and best-practice recommendations
LLM and GPU rental prices: model price lookup, GPU listings, cheapest-GPU search, price history
Benchmark your own prompt on 100+ LLMs with real API calls; pick the cheapest model that passes.
Routes each LLM completion to the cheapest capable provider across 8 providers (BYOK).
Intelligent LLM routing proxy — 60-90% cost savings by auto-selecting the cheapest model
LLM cost optimization. Audit your calls, prove savings, route to the cheapest quality model. BYOK.
Agentcents — questions
Answers built from our own checks of this server.