Audit codebases and apply safety-ordered fixes with user confirmation.
npx skills add https://github.com/thtskaran/claude-skills --skill deslop-thtskaran
Please help me install this Agent Skill. Skill: deslop Source: https://github.com/thtskaran/claude-skills/tree/main/deslop Command: npx skills add https://github.com/thtskaran/claude-skills --skill deslop-thtskaran
deslop helps you turn brittle, AI-generated, or “vibe-coded” projects into codebases that are safer, more resilient, and closer to production readiness by performing a structured audit and then applying fixes in a safety-ordered way.
Use deslop when you want to review, audit, harden, refactor safely, or fix quality issues across an existing codebase directory (you provide the path, and the skill produces AUDIT.md first, then waits for confirmation before fixes).
High-intent search queries and answers about installing and using this skill.
Production hardening for AI-generated code is achieved through a two-phase audit and fix workflow that evaluates error handling, security, and resilience. This process generates a structured AUDIT.md and then applies safety-ordered fixes while preserving intended business logic.
Auditing a fragile codebase involves reconstructing the codebase map, scanning patterns for consistency, tracing critical paths for resilience, and evaluating OWASP Top 10 style security issues. This produces machine-parseable findings structured in an AUDIT.md file.
The two-phase workflow enforces refactoring safety by strictly separating auditing from fixing. Phase 1 generates a structured AUDIT.md, and Phase 2 consumes that file to apply safety-ordered fixes. Phase 2 never starts without explicit user confirmation.
Fixes during a security audit are not applied automatically or immediately. The process enforces a strict two-phase workflow where Phase 2 applies fixes using a safety-tiered protocol, requiring user approval for higher-risk changes and characterization tests for intermediate tiers.
The audit is language-agnostic and stack-aware, meaning it reconstructs the codebase map and scans patterns regardless of the specific programming language. It systematically evaluates consistency, resilience, and security hygiene across diverse existing repositories.
Safety-tiered code hardening fixes require explicit guardrails that limit automation. Tier 3 changes mandate user approval, and Tier 2 changes require characterization tests, meaning higher-risk refactoring cannot proceed without passing these safety checkpoints.
harshitsinghbhandariExecute multi-stage refactoring plans from audit.md and improvements.md.
Corvalis-LLCAnalyze Git changes across nine dimensions and generate review artifacts.
telihaDiagnose and fix CI/CD pipeline failures across Foundry, Next.js, Node.js/TypeScript, and Rust projects.
eliottbusinessAutomate pre-commit code review with static analysis and security scanning.
