Trains SONA and MicroLoRA neural patterns from completed task trajectories.
npx skills add https://github.com/ruvnet/claude-flow --skill neural-train
Please help me install this Agent Skill. Skill: neural-train Source: https://github.com/ruvnet/claude-flow/tree/main/plugins/ruflo-intelligence/skills/neural-train Command: npx skills add https://github.com/ruvnet/claude-flow --skill neural-train
It captures what worked during successful task executions and converts those trajectories into reusable neural patterns, preventing learned knowledge from being lost between sessions.
Ask the agent to check neural status, train coordination patterns for 10 epochs from recent task completions, and then verify the stored patterns.
High-intent search queries and answers about installing and using this skill.
Record the task as a trajectory using the trajectory-start, trajectory-step, and trajectory-end hooks, then call neural_train with a pattern type such as coordination and an epoch count. Verify the result with neural_patterns to confirm the patterns were stored.
Use SONA for single-domain, real-time micro-adaptation under 0.05ms latency. Switch to MicroLoRA when you have three or more distinct domains, creating one adapter per domain so training in one domain does not overload the others.
The consolidate flag triggers EWC++ on weight deltas, which protects previously learned domain weights while applying new training. Run agentdb_consolidate after roughly every ten trajectory completions to fold patterns into long-term storage.
Call hooks_pretrain with a model type such as moe and an epoch count, then run hooks_build-agents with the agent types you need. Pretraining writes to the plural patterns namespace, distinct from the singular pattern ReasoningBank target.
Yes, call hooks_intelligence-reset to wipe all learned intelligence state. This is intended for testing and benchmarking only, since it permanently removes stored patterns and adaptations.
DNYoussefAutomate adaptive learning for AI agents using ReasoningBank and AgentDB.
agenticsorgCreate and train AI learning plugins with AgentDB reinforcement learning algorithms.
adibfirmanRun Go benchmarks with correct methodology and compare results using benchstat.
curiositechSelect optimal LLM models and providers based on task complexity and cost.
