# lidangzzz/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

Repository: https://github.com/lidangzzz/goal-driven
Canonical: https://ross.abutalabs.com/products/goal-driven
License Family: other
Last push: 2026-03-18T14:05:38+00:00

## Health v2 (maintenance only)
Score: 47/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 72, release rhythm 35, longevity 12
- inputs: {"age_days": 170, "days_push": 168, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, young, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1614, forks 125 (observed 2026-08-28T04:05:11.380804+00:00)

## What it is
Goal-Driven is a prompt-based multi-agent methodology that wraps coding agents like Claude Code or Codex in a master-agent loop, keeping subagents working for hundreds of hours on a single verifiable goal. The master agent periodically evaluates subagent output against explicit success criteria and restarts work until the criteria are met.

## Use cases
- run an AI agent for 100+ hours to build a compiler from specs
- have agents persistently solve a hard verifiable math or computational problem
- keep a coding agent from stopping early on a long complex task
- orchestrate master and subagent loops for goal completion checking
- generate large open-source projects like a SQLite clone with agents

## When to choose
- your task has a clear goal and strict machine-checkable success criteria
- you use a multi-agent coding tool like Claude Code, Codex, or OpenClaw
- the problem is complex, long-running, and verifiable such as compilers or theorem proving

## When to avoid
- your task is subjective or lacks well-defined completion criteria
- you need a quick one-shot coding assistant rather than a long autonomous run
- you cannot afford extended agent runtime and token costs

## Facets
- artifact type: framework
- maturity: experimental
- function: agent-framework, llm-inference, prompt-engineering, developer-tools
- domain: artificial-intelligence, large-language-models, developer-tools
- platform: cli, cross-platform
- tags: multi-agent, long-running-agents, goal-driven, claude-code, codex, autonomous-coding, verification-loop, ai-agents, automation

## Member repositories
- lidangzzz/goal-driven (main) score 47

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:11.380804+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T03:50:21.714812+00:00, confidence not recorded.
  - readme: https://github.com/lidangzzz/goal-driven (fetched 2026-08-28T04:05:11.380804+00:00, sha 5fd7b39c2427)
- Data as of 2026-08-30T08:39:29.467469+00:00.