project-claude-config

Auditing project-level Claude configurations requires identifying redundant hooks and settings that duplicate global defaults. This skill automatically critiques settings.json and CLAUDE.md files to ensure they remain lean, safe, and non-redundant for the specific repository environment.

Scaffolding a repository-specific CLAUDE.md file requires focusing on situational, project-specific instructions rather than generic best practices. This skill uses a single-pass, state-aware approach to generate concise instruction files optimized for agent context budget management.

Claude agent inconsistency often stems from configuration drift or duplicated instructions across global and project levels. This skill audits project settings and hooks to ensure deterministic enforcement and adherence to global safety guardrails, eliminating conflicting behaviors.

Managing the instruction budget requires scaffolding CLAUDE.md files to remain concise and focused on project-specific needs. This skill evaluates and critiques existing instructions to prevent context bloat and optimize agent behavior for the repository.

Yes, this configuration tool operates without external dependencies. It functions as a standalone skill to audit and scaffold project-specific settings.json and CLAUDE.md files, applying state-routed logic to optimize agent context and safety guardrails directly.

Automated scaffolding should be avoided when a repository requires highly custom, non-deterministic agent behaviors outside standard safety guardrails. This skill targets consistent, non-redundant agent behavior and deterministic hook enforcement for standard project-level setup.