Modulus: Cross-repository knowledge orchestration for AI coding agents
Modulus is a desktop application designed to orchestrate multiple AI coding agents working across multiple repositories with shared context. The tool addresses two specific problems developers face when using AI coding agents: broken cross-repository context and lost context when switching between agents.
Core functionality
The app enables running multiple AI coding agents simultaneously without conflicts, allowing developers to fix bugs while building features. Each agent operates in its own isolated workspace using git worktrees, preventing conflicts and eliminating the need for multiple IDE windows or repository cloning.
Key features
- Shared project memory: Agents automatically know your API schemas, dependencies, and recent changes across all repositories, eliminating the need for copying and pasting between windows
- Cross-repository understanding: Agents understand dependencies between repositories (e.g., backend repo + client repo + shared library repo + AI agents repo)
- Parallel agent execution: Run multiple coding agents working in parallel with shared context
- Review and ship workflow: Review all changes from all agents in one place and create pull requests directly from Modulus
- Memory and Context Engine: Built from the ground up specifically for coding agents
Technical implementation
Modulus connects to Cursor via MCP (Model Context Protocol). The tool uses git worktrees to create isolated workspaces for each agent. The team built their own Memory and Context Engine specifically for coding agent workflows.
Current availability and future plans
Currently available for macOS with a waitlist for Linux and Windows. The team plans to add team collaboration features allowing teams to share knowledge with others to improve workflows with AI coding agents. An upcoming API will allow developers to switch between coding agents or IDEs without losing context.
Use case example
The founders built Modulus to solve their own problem: while working across two different repositories, they had to manually paste API schemas between Cursor windows, repeatedly telling the frontend agent what the backend API looked like. Modulus eliminates this manual context sharing.
📖 Read the full source: HN AI Agents
👀 See Also
Skillware adds synthetic data generator with entropy scoring for local model fine-tuning
Skillware has released a new synthetic data generator skill that uses zlib compression-ratio heuristics to score output diversity, helping prevent model collapse. The tool works out-of-the-box with Ollama, supports Gemini/Anthropic for high-reasoning batches, and outputs JSON batches for .jsonl fine-tuning pipelines.
Buyer Eval: Claude skill for B2B vendor evaluation using AI agent conversations
A Claude skill that evaluates B2B software vendors by researching your company, asking domain-specific questions, and directly interrogating vendor AI agents through the Salespeak Frontdoor API. It cross-references claims against independent sources and produces evidence-based scorecards with transparent verification levels.
Open-Source Framework Uses Claude Code CLI for Automated GitHub Repo Monitoring
A developer has open-sourced a framework that runs Claude Code CLI on a cron schedule to triage GitHub activity across multiple repositories. The tool includes state tracking, deduplication, Discord notifications, and a pre-check system that avoids API costs when nothing has changed.
Browser39: A Headless Web Browser for AI Agents
Browser39 is a headless web browser designed specifically for AI agents that converts web pages to token-optimized Markdown locally, runs JavaScript, manages cookies and sessions, queries the DOM, and fills forms. It's a single binary with no external browser needed, no fees, and no external service.