Allocate research effort across ideas using Thompson-sampling and Laplace pseudo-counts.
npx skills add https://github.com/fkguo/nullius --skill idea-allocation
Please help me install this Agent Skill. Skill: idea-allocation Source: https://github.com/fkguo/nullius/tree/main/skills/idea-allocation Command: npx skills add https://github.com/fkguo/nullius --skill idea-allocation
This Skill streamlines the process of allocating research effort across a portfolio of ideas and monitoring their activation conditions, saving researchers time and ensuring efficient use of resources.
Use the idea-allocation skill to perform a budgeted allocation for the current research cycle. Provide the path to the campaign store snapshot and specify the seed, number of deep slots, and reconnaissance slots.
High-intent search queries and answers about installing and using this skill.
You can automate research portfolio allocation by applying Thompson-sampling and Laplace pseudo-counts to distribute effort across ideas based on posterior probabilities. This ensures evidence-based resource allocation for your research campaigns.
Monitor activation conditions to track research ideas waiting for external triggers. This process identifies ideas missing evidence and verifies if external dependencies are met for the current allocation cycle.
Thompson-sampling allocates resources by using posterior probabilities and Laplace pseudo-counts to balance exploration and exploitation. It dynamically assigns deep and reconnaissance slots to ideas based on their evidence and potential.
Yes, you need a campaign store and a belief graph to perform budgeted allocation. You must provide the campaign store snapshot, seed, and specify the number of deep and reconnaissance slots for the research cycle.
Yes, use this approach for decision-making when evidence is missing. The activation monitoring feature tracks ideas waiting on missing evidence and manages their status until external conditions are satisfied.
kriscendobotCoordinate decentralized job allocation among autonomous agents using reputation-weighted Thompson sampling.
yudameAutomate Perplexity Deep Research API calls for academic literature reviews.
melikhanmutluOrchestrate multi-actor AI workflows with graph-based state management.
lyndonklConvert probabilistic forecasts into betting decisions using Kelly Criterion and edge calculation.
