goal-driven
A multi-agent system that keeps running for ~100 hours and solve a very complicated coding or math problem that can be verified
Goal-Driven is a multi-agent orchestration framework that enables AI coding assistants to work continuously for 100+ hours on complex, verifiable problems. It uses a master agent to supervise subagents, repeatedly checking whether success criteria are met and restarting work when necessary. The system is designed for intricate tasks like compiler development, mathematical theorem proving, and database architecture that require sustained effort and can be objectively validated. Developers integrate it by providing a goal and success criteria as a prompt template to multi-agent tools.
Key Features
Use Cases
- 01Building a compiler from scratch by specifying the source and target languages as the goal
- 02Proving mathematical theorems where success is verified by formal proof checkers
- 03Implementing complex database systems with test suites that validate correctness
- 04Designing EDA simulation tools with comprehensive verification through test cases
- 05Solving computational problems that require extensive iteration and validation
- 06Creating system-level architectures where success criteria can be objectively measured
Related Agents
View morehermes-agent
The agent that grows with you
agency-agents
A complete AI agency at your fingertips - From frontend wizards to Reddit community ninjas, from whimsy injectors to reality checkers. Each agent is a specialized expert with personality, processes, and proven deliverables.
openinterpreter
A coding agent for open models like Kimi K3
cline
Autonomous coding agent as an SDK, IDE extension, or CLI assistant.
goal-driven — FAQ
What is the Goal-Driven AI sub-agent framework?+
Goal-Driven is a multi-agent orchestration system that keeps AI coding assistants working continuously on complex problems for 100+ hours. It uses a master agent to supervise subagents and verify that work meets strict success criteria before halting.
How do I use Goal-Driven with my AI agent?+
Copy the provided prompt template, fill in your specific goal and success criteria in the designated brackets, and paste the completed prompt into a multi-agent tool like Claude Code, Codex, or OpenClaw. The master agent will then create and manage subagents automatically.
Which AI clients work with Goal-Driven?+
Goal-Driven works with any multi-agent–supported AI tool, including Claude Code, Codex, and OpenClaw. The framework operates through prompt instructions rather than requiring specific software integration.
Does Goal-Driven require API keys or specific prerequisites?+
You need access to a multi-agent AI tool with sufficient API credits or subscription balance, as Goal-Driven processes can run for 100+ hours and consume significant LLM tokens. No additional API keys beyond your AI agent platform are required.
Is Goal-Driven free to use?+
The Goal-Driven framework itself is open source and free. However, running it will incur costs from your AI agent provider based on token usage and API calls during the extended execution time.
What kinds of problems is Goal-Driven best suited for?+
Goal-Driven excels at complex, time-consuming tasks with clear verification criteria, such as compiler development, mathematical theorem proving, database implementation, and system-level design challenges. The problem must be objectively verifiable for the master agent to determine success.
How do I install goal-driven?+
Open the source repository on GitHub and follow its README. goal-driven is a agent — MCP Agents Market links you directly to the official repo.
Is goal-driven free?+
goal-driven is an open-source project hosted on GitHub. Check the repository for its license and any usage requirements.