harness-kit CLAUDE.md

harness-kit CLAUDE.md is an instructions file for coding agents from deepklarity/harness-kit. It costs 7,165 tokens per session, scanned A, original, MIT.

Repository instructions for Claude Code, an AI coding assistant, covering project memory, Git safety, design principles, and how the agent should communicate with the operator.

In plain words
What is it for?
Following repository rules, storing project learnings, checking changes, and coordinating work across agents.
Why use it?
They keep shared work safe when several agents work at once and give agents one consistent place for project context. They also prevent unsafe use of Git stashes and misplaced memory files.

Instructions file

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add instructions/deepklarity/harness-kit/claude-md
Clone the repo
git clone --depth 1 https://github.com/deepklarity/harness-kit

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for harness-kit CLAUDE.md

README.md
[![agentmods](https://agentmods.dev/badge/instructions/deepklarity/harness-kit/claude-md.svg)](https://agentmods.dev/instructions/deepklarity/harness-kit/claude-md)
Your own site
<a href="https://agentmods.dev/instructions/deepklarity/harness-kit/claude-md"><img src="https://agentmods.dev/badge/instructions/deepklarity/harness-kit/claude-md.svg" alt="Measured on agentmods" height="20"></a>
Per session 7,165 This file is loaded in full into every session.
When invoked 7,165 The same file — it is already loaded in full.
Security scan A 1 finding. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.07165 $0.07165
Opus 5 $0.03583 $0.03583
Sonnet 5 $0.01433 $0.01433
Haiku 4.5 $0.00717 $0.00717

Measured 4d ago against content hash 4263b0ec8c42, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

harness-kit CLAUDE.md scanned grade A with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 4d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

**Never use `curl`, raw API calls, or direct HTTP requests** to inspect TaskIt data. The API requires Firebase auth. **Always use the diagnostic scripts** in `taskit/taskit-backend/testing_tools/` — they use Django ORM d
CLAUDE.md · 420 lines

How it starts

The opening of the file, as written. The whole thing — 420 lines — stays where its author put it; the contents beside it link to each section on GitHub.

CLAUDE.md

This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

Memory Policy

Never use MEMORY.md files. Do not create, read, update, or reference any MEMORY.md files in this repository. All persistent project context belongs in CLAUDE.md files (root and per-project). If you need to store learnings, use /hk-compound to write to docs/patterns/ instead.

Git Safety: No Stash

Never run ANY git stash command in the main repo checkout or any shared context. Multiple agents work in parallel on this repo; stash operations there create invisible shared state that causes silent data loss across concurrent sessions. Use branches, temporary commits, or disk-anchored files instead; check for uncommitted changes with git status or git diff.

Exception — isolated task worktrees: an agent running inside its own task worktree (.odin/worktrees/...) may use stash on its own work; the stash is scoped to that worktree and cannot affect other sessions. Prefer temporary commits anyway (a crashed run loses stashed work), but stash use inside an isolated worktree is NOT a rule violation and must not, on its own, fail a review.

Design Principles

Simplicity is a feature. Between two correct approaches, pick the one with fewer moving parts. The goal is code so straightforward it looks obvious — that's the hardest thing to achieve, not the easiest. Extra abstraction layers, wrapper functions, and extensibility points are costs that need justification, not defaults to reach for. When a shortcut is the right call, take it — but you should be able to say in one sentence why it won't cause problems later. If you can't, do it properly.

Default First: Every feature should work out of the box with sensible defaults. Configuration, agent assignments, and settings should be suggestive — provide a good default that the user can override, rather than requiring upfront configuration. The system picks reasonable choices automatically; the user intervenes only when they want something different.

Read the full file on GitHub · 420 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 4d ago First seen · 420 lines · 7,165 tokens per session scan A 4263b0ec8c42

Subscribe to this mod's changes

harness-kit CLAUDE.md is an instructions file published in the GitHub repository deepklarity/harness-kit (95 stars, last pushed 1mo ago), licensed MIT. It adds 7,165 tokens to every session, about $0.0358 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.