Identify capability gaps and execute evidence-based toolkit improvement cycles.
npx skills add https://github.com/notque/vexjoy-agent --skill toolkit-evolution
Please help me install this Agent Skill. Skill: toolkit-evolution Source: https://github.com/notque/vexjoy-agent/tree/main/skills/meta/toolkit-evolution Command: npx skills add https://github.com/notque/vexjoy-agent --skill toolkit-evolution
Prevents the toolkit from stagnating or accumulating weak, unmeasured workflow gaps by running a structured, evidence-based self-improvement loop.
Invoke toolkit self-improvement by running /evolve --discover to find missing skills and evidence-backed improvement opportunities.
High-intent search queries and answers about installing and using this skill.
Evidence-based improvement requires concrete signals like routing failures, git churn, and learning database records to block speculative changes. This ensures that proposed updates to skills and routing are validated through deterministic checks and A/B testing before promotion.
You run a self-improvement cycle by executing a discovery process to find missing skills, diagnosing root causes from repo history, building implementations on isolated feature branches, and validating outcomes with A/B testing before promoting via PR.
Yes, you can validate new features by implementing top winners on isolated feature branches, running deterministic validation checks, and comparing outcomes against a baseline. This A/B testing ensures only proven updates get promoted to the main toolkit.
Yes, the process requires concrete evidence such as routing failures, git churn signals, and learning database insights for every proposal. This evidence-gated approach blocks speculative improvements and ensures updates are driven by measured workflow gaps.
Automated toolkit evolution is limited by its reliance on concrete evidence; it blocks speculative improvements. It also requires multi-persona critique or a defined fallback, isolated feature-branch builds, and full cycle learnings recorded to prevent unmeasured workflow gaps.
mkalkereReproduce, isolate, and diagnose code failures with evidence-based techniques.
encero-systemsRecord and organize codebase learnings into categorized sections for future reference.
Julian0444Create isolated Git worktrees for feature development and multi-branch work.
cucurb-itDocument domain context, scope, and terminology for feature analysis requests.
