Audit Claude Code configuration and Python/ML projects, then generate repair and evolution plans.
npx skills add https://github.com/mykcs/myk-skills --skill rich-audit
Please help me install this Agent Skill. Skill: rich-audit Source: https://github.com/mykcs/myk-skills/tree/main/rich-audit Command: npx skills add https://github.com/mykcs/myk-skills --skill rich-audit
rich-audit eliminates configuration drift and quality issues in your Claude Code + memory ecosystem by running a structured audit, applying safe repairs, and then proactively evolving your setup with external best practices.
Ask it to run rich审计 to produce a full audit report, apply safe mechanical repairs, and generate an evolution plan with external search evidence.
High-intent search queries and answers about installing and using this skill.
A Claude Code configuration audit detects drift by running deterministic mechanical checks across rules, memory indexes, skills/symlinks, and hooks, generating structured JSON reports with severity scores and remediation action items.
Fixing MEMORY.md inconsistencies requires memory alignment verification to detect phantom, missing, or mem0 gaps between the memory file and the case-file filesystem, applying safe optional auto-fixes and generating case templates when needed.
Yes, dual-mode auditing supports checking Python and ML project health alongside Claude Code configuration to catch dependency risks, version mismatches, and completeness gaps within the same pipeline.
The best way to repair and evolve Claude Code rules and hooks is a three-layer pipeline that applies targeted safe auto-fixes after auditing, then generates an evolution plan using fresh external references and parallel agent orchestration.
Symlink consistency checking itself relies on deterministic mechanical checks, but the audit pipeline enforces mandatory post-audit external scanning to gather best practices and generate a safe evolution plan for your setup.
nishide-devValidates ML project structure, Hydra configs, and runtime instantiation paths before training runs.
mattstrubleEnforce a test-first workflow with feature, refactor, and bug fix modes.
Cortexa-LLCDelegate independent subtasks to specialized agents for parallel execution.
diShine-digital-agencyRe-run SHINE SessionStart hooks to refresh memory symlink, plugin sync, and update checks.
