AccInt is a local Work Model and MCP server for persistent scored agent memory, retrieval, commitments, browser/runtime observations, and outcome-based learning across coding agents.
Overview
AccInt
AccInt is a local Work Model for coding agents: scored memory, recursive retrieval, commitments, runtime/browser observations, and outcome feedback in one substrate. It exposes an MCP interface so agents can retrieve prior work, record decisions, and close the loop when reality validates or rejects an action.
Repository: https://github.com/maxbaluev/accreted-intelligence Website: https://accint.xyz
Server Config
{
"mcpServers": {
"accint": {
"command": "acc",
"args": [
"mcp"
]
}
}
}