harness AGENTS.md

A set of project instructions for Harness, a repository that stores Claude Code and Codex setup files, skills, safety hooks, schedules, and related installation scripts.

In plain words
What is it for?
Use it when maintaining Harness, installing its skills, updating the Codex adapter, configuring agents, or following its repository-specific constraints.
Why use it?
It documents how to install or refresh shared skills and agent settings while keeping the repository as the source of truth. It also explains the separate Codex adapter and custom-agent setup.

Instructions file for CodexOpenCode

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/ryan-scheinberg/harness/agents-md
Clone the repo
git clone --depth 1 https://github.com/ryan-scheinberg/harness

Made for: Codex, OpenCode.

Per session 2,162 This file is loaded in full into every session.
When invoked 2,162 The same file — it is already loaded in full.
Security scan B 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 $0.02162 $0.02162
Opus 5 $0.01081 $0.01081
Sonnet 5 $0.00432 $0.00432
Haiku 4.5 $0.00216 $0.00216

Measured 2d ago against content hash 1becb9d3ef76, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade B, and why

harness AGENTS.md scanned grade B 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 2d 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.

Reads agent configuration directoriesmediumAgent snooping

.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.

| `./codex-agents/install.sh` | Installs native Codex custom-agent TOML into `~/.codex/agents/` and a marked MultiAgent V2 block into `~/.codex/config.toml`. The block exposes custom role/model selection and sets eight t
AGENTS.md · 80 lines

How it starts

The opening of the file, as written. The whole thing — 80 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Harness repo — agent notes

This repository is the source of truth for the user's Claude Code harness. Skills, the global CLAUDE.md, safety hooks, local schedules, and the optional Codex adapter all live here and are version-controlled in one place.

Install / refresh

Command What it does
./skills/install.sh Discovers every SKILL.md under skills/ (any depth, skipping .git/) and flat-symlinks each directory into ~/.claude/skills/<name>/. Prunes stale symlinks pointing into $REPO_ROOT or legacy ~/Documents/skills/. You can run this
./codex/install.sh Optional Codex adapter. Installs shared skills into ~/.codex/skills/<name>/; non-role skills are symlinks, role skills are generated copies with Codex $role-* invocation wording so Claude role files stay untouched. Symlinks ~/.codex/AGENTS.md to the separate ignored personal codex/AGENTS.md; it never reads CLAUDE.md. You can run this
./codex-agents/install.sh Installs native Codex custom-agent TOML into ~/.codex/agents/ and a marked MultiAgent V2 block into ~/.codex/config.toml. The block exposes custom role/model selection and sets eight total V2 slots; unrelated config stays untouched. You can run this
./hooks/install.sh Merges safety hooks (native deny rules + the PreToolUse bash gate and codex-timeout hooks) into ~/.claude/settings.json. Writes to user settings. The user will run this
./schedules/install.sh Syncs schedules/*.cron specs into the user's crontab under a delimited block. Writes to crontab, and needs macOS Full Disk Access. The user will run this

CLAUDE.md is a one-time manual symlink: ln -sf "$PWD/CLAUDE.md" ~/.claude/CLAUDE.md

Installers are Bash 3.2-compatible so they run on macOS's default shell. Re-run an installer after adding, moving, or renaming anything in its directory. The repo is the source of truth: installers unconditionally overwrite their destinations, no adoption check or diff dance

Read the full file on GitHub · 80 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. 2d ago First seen · 80 lines · 2,162 tokens per session scan B 1becb9d3ef76

Subscribe to this mod's changes

harness AGENTS.md is an instructions file published in the GitHub repository ryan-scheinberg/harness (2 stars, last pushed 1mo ago), licensed MIT. It adds 2,162 tokens to every session, about $0.0108 per session on Opus 5. A static security scan graded it B with 1 finding (reads agent configuration directories). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.