Research Pipeline FAQs

Question 1

Can I resume the pipeline if it fails or is interrupted?

Accepted Answer

Yes. The pipeline persists its state to a session directory in '.omc/research-pipeline/', allowing it to track progress and resume from the last completed stage.

Question 2

Which Claude models are used during execution?

Accepted Answer

The pipeline uses a multi-tier approach: Opus-tier agents are used for deep analysis and creative synthesis, while Sonnet-tier agents handle standard summarization and web searching tasks.

Question 3

What is the output of the final stage?

Accepted Answer

The final stage (Stage 8) produces a comprehensive experiment plan that includes baseline comparisons, evaluation metrics, and the design for testing the synthesized proposal.

Question 4

Does this skill only analyze the papers I provide?

Accepted Answer

No. In Stage 2, the pipeline identifies gaps in your current literature and uses WebSearch to find and automatically download compensatory papers into your directory for analysis.

Question 5

How do I specify a baseline paper for the pipeline?

Accepted Answer

You can identify a baseline paper by prefixing its filename with 'baseline_' (e.g., baseline_paper.pdf) or by placing it in a subdirectory named 'baseline/'. If neither is found, the skill will ask you to specify one.

Research Pipeline

主要功能

使用场景

Research Pipeline

主要功能

使用场景