Harness Engineering

muratcankoylan/agent-skills-for-context-engineering

You need this when you're building agents that run for hours or days without supervision. It covers the control plane: which files the agent can edit versus which stay locked (like eval metrics), how to structure append-only logs so crashes don't lose work, when to gate on novelty checks, and where to draw the human approval line. The core insight is that feedback loops break down when agents can modify their own success criteria, so you separate editable surfaces from locked rubrics. If you're setting up research loops, PR-generating workflows, or anything that needs rollback and durable state, this gives you the checklist and file patterns to keep autonomy from drifting into chaos.

skills.sh installs: 189
GitHub stars: 18.3K
last updated: May 26, 2026

View on GitHub

Install

$ npx -y skills add muratcankoylan/agent-skills-for-context-engineering --skill harness-engineering --agent claude-code

Installs into .claude/skills of the current project.

Files

Harness Engineering

Harness engineering designs the control system around an agent: what it may edit, how it receives feedback, where it writes state, how failures recover, and who can approve irreversible actions. The harness is the difference between a helpful agent session and an autonomous loop that can run for days without corrupting its objective.

When to Activate

Activate this skill when:

Building autonomous research or experimentation loops
Designing an agent environment with locked metrics and editable code or content
Creating PR-producing or background agents
Evaluating whether an agent can safely run without frequent human prompts
Adding novelty, ablation, pruning, rollback, or durable logging to an agent workflow
Preventing agents from gaming benchmarks, weakening rubrics, or losing state across compaction

SKILL.md on GitHub