research-pipeline

Automating the research pipeline involves orchestrating idea discovery, implementation, experiment deployment, and an auto-review loop to generate findings. This Skill executes the full research lifecycle autonomously, optionally generating a paper for a specified venue like ICLR.

An auto-review loop in research automation iterates on experimental results and implements fixes automatically. This Skill uses the loop to continuously validate research outcomes, refining the implementation until the experimental design meets the required standards.

Yes, autonomous research workflows require WebSearch, Bash, and other tools like Read, Write, and Edit to deploy experiments. This Skill leverages these dependencies alongside codex services to automate experimental design and execution.

Start end-to-end research automation by invoking the pipeline with a target topic and parameters like difficulty and auto_write. Use the command structure to set human checkpoints to false and specify the target venue for paper generation.

Automating paper writing depends on the quality of generated experimental results and the auto-review loop. This Skill optionally generates papers for venues like ICLR, but setting human checkpoint to false means the review process is fully automated without manual validation.

Yes, codex services are required to run this automated experiment deployment pipeline. The Skill depends on mcp__codex__codex and mcp__codex__codex-reply alongside standard tools like Bash and WebFetch for comprehensive research automation.