Cursourcing

Codex plans. Cursor executes. Codex reviews.

4 followers

Codex plans. Cursor executes. Codex reviews.

4 followers

Cursourcing lets Codex delegate scoped coding tasks to Cursor CLI, then bring results back for review. Keep GPT-6 Astra focused on planning and judgment while Cursor handles implementation, investigations, and follow-ups.

Interactive

Launch tags:Open Source•Developer Tools•Artificial Intelligence

Launch Team / Built With

Framer AI AgentsDesign and publish professional sites with AI

Promoted

OpenAI

GPT-6 Astra Challenge

What became possible in your product with Astra that was not practical before?

Maker

Before GPT-6 Astra, Cursourcing was an idea that required manually coordinating two separate coding agents. Astra made it practical to turn a long-horizon goal into scoped subtasks, decide what Codex should keep, delegate implementation to Cursor, track progress, recover sessions, and review results. This lets us use Codex for planning and judgment while Cursor executes delegated work in parallel, without forcing developers to replace their existing tools. The key change is not just better code generation; it is making agent-to-agent delegation reliable enough for a real development workflow.

Report

1mo ago

Maker

📌

Hi Product Hunt! We’re Cogine AI, the team behind Cursourcing. We built Cursourcing around a simple problem: when Codex is working on a long task, useful execution capacity in Cursor can sit idle. Cursourcing lets Codex choose a scoped subtask, hand it to Cursor CLI, track progress, answer questions, and review the result before bringing it back into the main task. The workflow stays explicit: Codex plans and makes decisions. Cursor executes the delegated work. Codex reviews the result. It supports implementation tasks, investigations, parallel work, cancellations, session recovery, and follow-ups. Each tool keeps its own account and usage allowance. For the GPT-6 Astra Challenge, we used GPT-6 Astra in Codex for architecture, task decomposition, implementation review, testing, documentation, and release preparation. The result is an open-source Codex plugin that connects Codex and Cursor without replacing either workflow. The plugin is Apache-2.0 licensed, and the marketplace release includes a bundled runtime, so users do not need to clone the repository or run a separate build step. Install it with: codex plugin marketplace add cogine-ai/marketplace codex plugin add cursourcing@cogine-ai We’d love feedback from developers: where would you trust cross-agent delegation, what should be visible during a handoff, and which workflows should we support next?

Report

1mo ago

Reviews

OpenAI

GPT-6 Astra Challenge

What became possible in your product with Astra that was not practical before?

Maker

Report

1mo ago

Maker

📌

Report

1mo ago