Detect and repair AI-introduced code slop over a specified duration.
npx skills add https://github.com/av/harbor --skill anneal
Please help me install this Agent Skill. Skill: anneal Source: https://github.com/av/harbor/tree/main/.agents/skills/anneal Command: npx skills add https://github.com/av/harbor --skill anneal
The anneal Skill unit addresses the issue of AI-introduced architectural decay in codebases, such as duplicated logic, over-engineering, and silent error swallowing, by systematically fixing these issues over a user-defined duration.
Run the 'anneal' skill on your codebase for 4 hours to systematically fix AI-introduced code slop.
High-intent search queries and answers about installing and using this skill.
To fix AI-introduced code slop, you can run a systematic process that detects and repairs architectural decay like duplicated logic and over-engineering over a user-specified duration. The process uses subagents to perform individual fixes, run tests, and commit changes.
AI code slop is architectural decay introduced by AI, including duplicated logic, over-engineering, and silent error swallowing. It degrades code maintenance by creating hidden technical debt that accumulates silently during automated generation.
You can safely remove over-engineered code by following a strict safety protocol that makes one focused change at a time, runs tests, and reverts on failure. This ensures no changes occur to public APIs or external behavior.
Yes, you can automate silent error swallowing detection and repair by running a systematic analysis over a defined duration. Subagents identify these specific code slop patterns, apply fixes, run tests, and commit the changes automatically.
The best way to run systematic code optimization over several hours is to specify a user-defined duration for the process. Subagents will continuously detect and repair code slop patterns, verifying every change with tests before committing.
A key limitation when fixing architectural decay is that the process is bound by your specified duration and strict safety protocols. It makes one focused change at a time and reverts on test failure, ensuring no public API or external behavior is altered.
kynopticRun, diagnose, and fix failing test suites across language ecosystems.
Jamie-CuiFinalize feature branches with pre-integration checks and merge strategy guidance.
ShankulkarniDetects AI-generated slop patterns in JavaScript/TypeScript codebases.
vtgiang-dotcomSimplify code readability by replacing nested logic with guard clauses and named helpers.
