harness-gc

Audit documentation, feature integrity, and architectural drift to repair harness entropy.

1|Updated Feb 11, 2026
One-click install
npx skills add https://github.com/LauraFlorentin/skills-marketplace --skill harness-gc
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: harness-gc
Source: https://github.com/LauraFlorentin/skills-marketplace/tree/main/harness-engineer/skills/harness-gc
Command: npx skills add https://github.com/LauraFlorentin/skills-marketplace --skill harness-gc

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

Harness entropy accumulates as documentation, feature states, and architecture drift diverge from reality. This skill centrally coordinates automated cleanup to restore consistency, reduce cognitive load for engineers, and keep agents properly oriented by generating prioritized remediation reports and applying agent-led fixes.

Core Features & Use Cases

  • Documentation health checks: AGENTS.md references and docs alignment are verified and repaired where possible.
  • Feature integrity & progress: detects stalled in_progress features and broken circuit states, and suggests or applies resets.
  • Architecture discipline: identifies layer violations and dead context, enabling safer refactors.
  • Automated cleanup workflow: produces a prioritized cleanup plan and executes fixes across the harness over time.

Quick Start

Run the harness-gc check to scan the repository and produce a cleanup plan for the harness.

Frequently Asked Questions about harness-gc

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I fix AGENTS.md documentation drift and stale references in my repository?

To fix documentation drift, you can run an automated harness cleanup check to audit AGENTS.md references and docs alignment. This process identifies dead references and generates a prioritized remediation plan to repair inconsistencies across the repository.

What is the best way to detect stalled in_progress features and broken circuit states?

Detecting stalled in_progress features requires auditing feature integrity through features.json. This automated cleanup process identifies broken circuit states and suggests or applies resets to restore healthy execution progress.

How do I identify architecture layer violations and dead context in agent-first projects?

Identifying architecture layer violations involves auditing architectural drift across the harness. This cleanup action detects dead context and layer violations, enabling safer refactors and maintaining architectural discipline.

Can I automate repository cleanup for documentation, features, and architecture in one workflow?

Yes, you can automate repository cleanup by running a harness-gc check that scans AGENTS.md, docs/, and features.json together. It produces a prioritized cleanup plan and executes agent-led fixes across the harness over time.

Does harness cleanup work for typical agent-first projects without external dependencies?

Harness cleanup works for typical agent-first projects without requiring external dependencies. It centrally coordinates automated cleanup by implementing checks, resets, and archival actions that satisfy safety and reproducibility requirements.

When should I run an automated harness cleanup to repair entropy in my repository?

You should run harness cleanup when entropy accumulates as documentation, feature states, and architecture drift diverge from reality. It restores consistency, reduces cognitive load, and keeps agents properly oriented through prioritized remediation reports.