Hivelore is an open-source agent harness and contextual memory orchestrator created by Doucs91. It provides persistent project memory, decision tracking, and contextual briefing mechanics to enhance AI coding agents operating across CLI, MCP, and IDE environments. In day-to-day development, Hivelore runs as a CLI utility and Model Context Protocol (MCP) server that connects local repository context with agents like Claude, Cursor, and Roo Code. It analyzes repository churn, anchor specificity, and architectural decisions, automatically serving structured context through MCP bridges or rulesets stored in `.cursor/rules`, `.roo/rules`, and `.ai` directories. What differentiates Hivelore is its focus on long-term decision memory and git-aware contextual briefing. Instead of dumping raw file trees or naive embeddings, Hivelore manages memory promotion and deletion cycles (`memory approve/delete`) alongside git-history-based churn scoring to provide high-signal context to coding models. Hivelore is distributed as open-source software on GitHub under active versioned releases, supporting local execution and MCP integrations without proprietary platform lock-in.
Tags: agent harness, ai agent, cli, coding agent, developer tools, OpenSource
hivelore-mcp service bridge. - Decision Memory Tracking — Stores and manages project decision records and pre-adoption rules inside structured .ai repositories. - Memory Lifecycle Management — Provides explicit commands (memory approve, memory delete) to curate and promote agent memories safely. - Git-Aware Briefing & Churn Analysis — Uses commit depth and file churn metrics (IDF over commit history) to calculate anchor specificity and identify active code paths. - Multi-Editor Ruleset Integration — Generates and maintains structured rules for developer environments including Cursor (.cursor/rules) and Roo Code (.roo/rules). - Doctor Diagnostics — Built-in diagnostics command that verifies environment integrity and provides automated fix hints for configuration drift..ai decision memory to ensure context on architecture patterns, API constraints, and release rules persists without re-prompting.hivelore-mcp to expose live memory curation and repository context dynamically during chat and task execution.Hivelore is designed for software engineers and teams orchestrating AI agents across CLI and MCP-enabled editors (such as Cursor, Roo Code, and Claude) who need structured repository memory. It is less suitable for users looking for an all-in-one standalone GUI agent without terminal or MCP configuration requirements.
bash # Clone the repository git clone https://github.com/Doucs91/hivelore.git cd hivelore # Install dependencies and build npm install npm run build # Link the CLI binary globally npm linkbash hivelore doctor To configure Hivelore as an MCP server in your MCP client config (e.g., Claude Desktop or Cursor/Roo): json { "mcpServers": { "hivelore": { "command": "hivelore-mcp" } } }bash # Check context briefings and churn stats hivelore briefing # Approve or curate new decision memory proposals hivelore memory approvePart of CLI Coding Agents
Categories: AI Coding Agents, Code Assistants, Developer Tools