Propagate monotone values across interdependent cells until fixpoint convergence.
npx skills add https://github.com/plurigrid/asi --skill propagators
Please help me install this Agent Skill. Skill: propagators Source: https://github.com/plurigrid/asi/tree/main/skills/propagators Command: npx skills add https://github.com/plurigrid/asi --skill propagators
Presents Sussman/Radul propagator networks for constraint propagation and bidirectional dataflow, enabling fixpoint computation without explicit control flow.
Create a small propagator network and run until no new information arises, following the patterns described in the SKILL.md.
High-intent search queries and answers about installing and using this skill.
Constraint propagators apply monotonic updates across interdependent cells until reaching a fixpoint. They watch cells, compute derived values, and push updates to neighbors automatically, eliminating explicit control flow. This enables bidirectional dataflow where information flows in both directions to resolve constraints.
Use propagators when you have networks of interdependent values that must merge information and converge automatically. They handle concurrent, mergeable, and backtracking behavior without manual orchestration, making them ideal for constraint systems, numeric intervals, and symbolic reasoning.
Propagators solve constraints over monotone join-semilattice values, including numeric addition, symbolic intervals, and modular arithmetic constraints like GF(3). Any constraint system where values only increase or merge through a combining operation fits the propagator model.
Understanding monotonic updates and fixpoint computation helps, but the propagator framework handles the mechanics. Start with simple numeric or interval constraints; the provided Python and Julia examples demonstrate patterns without requiring deep theoretical knowledge.
Propagators compute bidirectionally and reach fixpoint automatically without explicit scheduling. Traditional dependency graphs require manual topological sorting and control flow; propagators merge information from all directions until convergence, handling cycles and concurrent updates naturally.
Yes, propagators support concurrent, mergeable updates. New information automatically triggers recomputation and propagation to dependent cells, making them suitable for live constraint systems where values change dynamically.
plurigridTransfer knowledge structures between domains using local scoped propagator networks.
nahisahoRetrieve GWAS Catalog associations, traits, and study metadata via the NHGRI-EBI REST API.
rainoftimeEncode program properties into SMT formulas and invoke solvers like Z3 or CVC5.
GhostOf0daysExecute convex optimization workflows with diagnostics and production controls.
