What is this?
An Agent Skill that helps coding agents prove and safely remove accidental complexity from an existing codebase without changing its real behavior.
An Agent Skill that helps coding agents prove and safely remove accidental complexity from an existing codebase without changing its real behavior.Is this a standalone app or a skill I add to my coding agent?It is a SKILL.md-based Agent Skill, not a standalone app. You install it into an agent such as Codex so the agent itself can run the audit-and-cleanup workflow.
It is a SKILL.md-based Agent Skill, not a standalone app. You install it into an agent such as Codex so the agent itself can run the audit-and-cleanup workflow.Do I need Node.js for the visualization feature?Yes, the built-in Cleanup Map renderer needs Node.js 18 or newer, but it has no extra npm dependencies and the generated HTML report doesn't fetch external fonts.
Yes, the built-in Cleanup Map renderer needs Node.js 18 or newer, but it has no extra npm dependencies and the generated HTML report doesn't fetch external fonts.Does it always delete the code it flags?No. When a real consumer is found, the ownership boundary is unclear, or the removal would just push the complexity elsewhere, it recommends keeping the code instead of forcing a deletion.
No. When a real consumer is found, the ownership boundary is unclear, or the removal would just push the complexity elsewhere, it recommends keeping the code instead of forcing a deletion.Which coding agents can use this skill?It's designed for Codex, and any other agent environment that supports the SKILL.md format can drop this repo into its own Skill directory.
It's designed for Codex, and any other agent environment that supports the SKILL.md format can drop this repo into its own Skill directory.What's the difference between Survey and Change mode?Survey is a read-only audit (focused on one subsystem or broad across the whole repo) that only produces findings. Change is authorized work that proves and completes an actual simplification boundary, applied in batches by ownership and verified independently.
Survey is a read-only audit (focused on one subsystem or broad across the whole repo) that only produces findings. Change is authorized work that proves and completes an actual simplification boundary, applied in batches by ownership and verified independently.What does a proof record actually contain?For each candidate it documents the ownership boundary, symbol, file and line; what maintenance burden it adds; who the production, test, dynamic and external consumers are; the full deletion boundary; what behavior or compatibility would be given up; and the minimal check that would reveal a wrong removal.
For each candidate it documents the ownership boundary, symbol, file and line; what maintenance burden it adds; who the production, test, dynamic and external consumers are; the full deletion boundary; what behavior or compatibility would be given up; and the minimal check that would reveal a wrong removal.Has this been tested on a real large codebase?Yes, it has been validated through Change, Broad, Integration and Decision-record scenarios, including a full-repo audit on a 973-file Python + TypeScript project, with methodology and known limits documented in docs/validation.md.
Yes, it has been validated through Change, Broad, Integration and Decision-record scenarios, including a full-repo audit on a 973-file Python + TypeScript project, with methodology and known limits documented in docs/validation.md.Do I need to install the visualization tool Archify separately?No. The interactive Cleanup Map renderer, built on Archify's Architecture rendering and desktop viewer core, is already bundled inside this skill.
No. The interactive Cleanup Map renderer, built on Archify's Architecture rendering and desktop viewer core, is already bundled inside this skill.What are similar projects?reclaim-code-entropy, cyclomatic-complexity-skill, draw-architecture, shuohao-skills, repowise, drawio-skill, Coding-Agent-Skills, darwin-skill
reclaim-code-entropy, cyclomatic-complexity-skill, draw-architecture, shuohao-skills, repowise, drawio-skill, Coding-Agent-Skills, darwin-skillWhich topics is it in?refactoring, skill-md, code-quality, architecture, static-analysis, visualization, skill, html, coding-agent, agent-skills, codex, agent
refactoring, skill-md, code-quality, architecture, static-analysis, visualization, skill, html, coding-agent, agent-skills, codex, agent