What problem does it solve?
Default full ECC installs often include irrelevant components that add noise and context overhead to repository workflows, and hand-curating a project-specific ECC set is tedious and subjective. This Skill solves that by classifying all ECC components using concrete evidence from the actual codebase, ensuring only relevant components are loaded by default.
Core Features & Use Cases
- Evidence-Based Classification: Sorts ECC skills, commands, rules, hooks, scripts, and extras into DAILY (always loaded) and LIBRARY (searchable reference) buckets using repo-local data like file extensions, lockfiles, framework configs, and dependency manifests.
- Flexible Review Workflow: Supports both parallel subagent passes and sequential review for all ECC component types, adapting to available tooling.
- Complete Output Artifacts: Generates a full DAILY/LIBRARY inventory, actionable install plan, verification report, and optional searchable skill-library router for off-stack components.
- Use Case: For a TypeScript Next.js repository, this Skill will promote TypeScript rules and frontend workflow skills to DAILY, demote unused Python or Django components to LIBRARY, and produce a ready-to-apply install plan that matches the repo's actual stack.
Quick Start
Use the agent-sort skill to analyze the current repository's codebase and generate an evidence-based ECC install plan that separates daily workflow components from library reference components.