Compare two research ideas via judged votes and generate a pairwise_match_v1 artifact.
npx skills add https://github.com/fkguo/nullius --skill idea-pairwise-match
Please help me install this Agent Skill. Skill: idea-pairwise-match Source: https://github.com/fkguo/nullius/tree/main/skills/idea-pairwise-match Command: npx skills add https://github.com/fkguo/nullius --skill idea-pairwise-match
This Skill enables pairwise comparison between two research ideas in a campaign, providing a relative-merit signal for decision-making.
pairwise_match_v1 artifact documenting the comparison and outcome.Run the pairwise comparison for ideas 'node_A' and 'node_B' with the following command: idea-pairwise-match node_A node_B --campaign-id <campaign_id>.
High-intent search queries and answers about installing and using this skill.
Run pairwise comparison of two research ideas by executing the match command with node identifiers and a campaign ID, collecting independent votes from a diverse judge panel to generate a documented outcome artifact.
A judged panel for idea comparison collects independent votes from a diverse set of judges representing different model families to evaluate paired research ideas against committed criteria within a campaign.
To compare two research ideas for advancement, run a pairwise match using the idea nodes and campaign ID, which triggers a judged panel to vote and outputs a pairwise_match_v1 artifact recording the comparison.
Yes, a commitment to specific evaluation criteria and prepared research materials are required prerequisites before initiating the pairwise comparison process within a campaign.
Use pairwise matching when a campaign needs to compare exactly two ideas for relative merit advancement, or when a new idea requires initial calibration against an existing benchmark idea.
Judging paired research ideas generates a pairwise_match_v1 artifact that documents the committed criteria, the collected judge panel votes, and the final comparison outcome.
Synaptic-Labs-AIConduct structured research and documentation gathering for the PACT Prepare phase.
JuniYadiAggregate five judge verdicts into a final decision and next steps.
gauravb4875-uxResearch products, compare prices, and generate gift ideas from multiple sources.
Peter-Swain-IncScore vendors with a weighted MCDA framework across eight evaluation dimensions.