Claude Code Harness

Design a real harness for an agentic system using Claude Code-inspired patterns. Use when the user needs a harness blueprint, request assembly design, execution loop, tool runtime, memory layering, permission model, transcript or recovery strategy, or wants to turn a vague agent idea into a harness-level architecture. Do not use for generic product brainstorming, simple prompt writing, or isolated code generation.

13 stars 1 forksAI, LLMs & Agents Plugin marketplace

Install

Claude Code plugin
/plugin marketplace add dadwadw233/claude-code-harness

The repo is a Claude Code plugin marketplace. Run /plugin afterwards to pick the plugin that includes this skill.

skills CLI (Claude Code, Cursor, Codex and more)
npx skills add dadwadw233/claude-code-harness --skill claude-code-harness
Manual install
git clone --depth 1 https://github.com/dadwadw233/claude-code-harness.git /tmp/claude-code-harness
mkdir -p ~/.claude/skills
cp -r /tmp/claude-code-harness/skills/claude-code-harness ~/.claude/skills/claude-code-harness

Personal skills live in ~/.claude/skills. Use .claude/skills in a project folder to share a skill with your team.

From the SKILL.md

# Claude Code Harness

Design the harness, not just the agent prompt.

This skill helps turn an agent idea into a harness blueprint that specifies:

- how a turn is assembled
- how the execution loop runs
- how tools are governed
- how memory survives boundaries
- how permissions and recovery are enforced
- how the system stays understandable to humans

## What to optimize for

Prefer harnesses that are:

- explicit rather than magical
- layered rather than blended
- recoverable rather than one-shot
- bounded rather than overpowered
- product-useful rather than architecturally theatrical

## Output contract

When this skill is active, default to a harness blueprint with these sections:

1. System goal and delegation boundary
2. Harness layers
3. Request assembly design
4. Turn loop design
5. Tool and capability runtime
6. Context and memory model
7. Permissions and safety gates
8. Transcript and recovery model
9. Extension surfaces
10. Implementation sequence

If the user is reviewing an existing design, lead with structural weaknesses and missing control planes.

## Workflow

### 1. Clarify whether the user needs a harness at all

Start by answering:

- what job is being delegated
- what state must survive across steps or turns
- whether the system needs tools, memory, recovery, or background work
- what would break if the model were only called once

If the workflow does not need a harness, say so directly.

### 2. Define the harness boundary

Separate:

- the model
Read the full skill on GitHub

Details

Skill name
claude-code-harness
Path
skills/claude-code-harness/SKILL.md
License
MIT
Repo updated
5 April 2026

FAQ

What does the Claude Code Harness skill do?

Design a real harness for an agentic system using Claude Code-inspired patterns. Use when the user needs a harness blueprint, request assembly design, execution loop, tool runtime, memory layering, permission model, transcript or recovery strategy, or wants to turn a vague agent idea into a harness-level architecture. Do not use for generic product brainstorming, simple prompt writing, or isolated code generation.

How do I install the Claude Code Harness skill?

Claude Code plugin: /plugin marketplace add dadwadw233/claude-code-harness. Skills go in ~/.claude/skills (personal) or .claude/skills (project), and Claude loads them automatically when a task matches the description.

Is Claude Code Harness popular?

It lives in dadwadw233/claude-code-harness, which has 13 GitHub stars and 1 forks, and is included in 1 curated skill list.

Related skills

Event Finder

39

ShawhinT/non-coding-skills

Find upcoming in-person events on Eventbrite for Shaw to attend, ranked by how close they are to Frisco/Plano and by topic (AI first, then tech, then general business networking). Use whenever Shaw asks to find events, meetups, or networking opportunities, e.g. "find AI events in DFW", "what's happening on Eventbrite next month", "any good networking events near me", "events to go to this quarter", "where can I meet founders/CEOs in person", or names a city plus a topic and a time window. Also fires for prospecting-by-showing-up requests like "where should I be in person to meet my ICP" when t

AI, LLMs & Agents

Aba Plugin Updater

39

ShawhinT/non-coding-skills

Prepare a change to Shaw's aba-plugins marketplace repo (the public plugin clients install via /plugin install aba-skills@aba) as a handoff doc for Claude Code, which executes it natively in the repo. Use whenever he wants to edit, fix, or improve one of the plugin's skills (skill-helper, skill-updater, cost-optimizer), repackage dist/, validate the manifests, or land a commit there. Triggers: "update the plugin", "push a change to aba-plugins", "repackage dist", "cut a plugin release", "add this to the aba-skills plugin". Also fire whenever one of those three skill names comes up as something

AI, LLMs & Agents

Sync Desktop Skills

39

ShawhinT/non-coding-skills

Import or update skills in this public repo from Shaw's three private sources — the canonical repo at ~/agents/shaws-skills/skills (rsynced from Claude Desktop), the Code-native skills in ~/.claude/skills, and the aba-skills plugin clone — scrubbing PII on the way in. Use whenever Shaw says "sync skills", "update the non-coding skills", "pull <skill> from canonical", "add these skills to the repo", or otherwise references moving his private canonical skills into this public repo. Handles the deterministic scrub script, parallel sub-agent judgment passes, README table maintenance, and a clean c

AI, LLMs & Agents

Swiftdata

42

robbyt/claude-skills

This skill should be used when working with SwiftData — @Model definitions, @Query, @Relationship, ModelContext, ModelContainer, schema migration, Swift 6 concurrency, performance optimization, or architecture review. Also triggers on: "SwiftData audit", "@Model issues", "SwiftData crashes", "predicate problems", "background context patterns", "N+1 queries", "SwiftData vs SQLiteData vs GRDB", "CloudKit sync with SwiftData", "SwiftData migration fails", or "SwiftData migration diagnostics". Covers auditing existing code, writing new SwiftData code, migration debugging, and architecture review.

AI, LLMs & Agents

Vtex Io Masterdata Strategy

47

vtex/skills

Apply when deciding whether and how VTEX IO apps should use Master Data v2 for custom data. Covers entity boundaries, schema lifecycle, indexing strategy, and when Master Data is the right storage mechanism versus another data approach. Use for reviews, wishlists, forms, or other custom data modeling decisions in VTEX IO apps.

AI, LLMs & Agents

Reflect

42

robbyt/claude-skills

Analyze recent conversation to identify improvements for CLAUDE.md memory files. Trigger when user says "reflect on this session", "reflect on this conversation", "reflect on this code", or "improve CLAUDE.md by reflecting on this session". The verb "reflect" combined with a session/conversation/code subject is the key trigger.

AI, LLMs & Agents

Browse the most popular Claude skills, all ai, llms & agents skills, or write your own with the skills generator.