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.
npx agentmods add instructions/joshuashunk/harbor/claude-mdgit clone --depth 1 https://github.com/JoshuaShunk/HarborWrote 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.
[](https://agentmods.dev/instructions/joshuashunk/harbor/claude-md)<a href="https://agentmods.dev/instructions/joshuashunk/harbor/claude-md"><img src="https://agentmods.dev/badge/instructions/joshuashunk/harbor/claude-md.svg" alt="Measured on agentmods" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00624 | $0.00624 |
| Opus 5 | $0.00312 | $0.00312 |
| Sonnet 5 | $0.00125 | $0.00125 |
| Haiku 4.5 | $0.00062 | $0.00062 |
Grade A, and why
Harbor CLAUDE.md scanned grade A with 0 findings 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 5d 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.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 58 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Harbor
Universal MCP Hub — desktop app + CLI that manages MCP servers across Claude Code, Codex, VS Code, and Cursor.
Tech Stack
Rust + Tauri v2 + React 19 + TypeScript + Tailwind CSS + Axum
Project Structure
crates/harbor-core— Core library (config, connectors, gateway, vault, marketplace, fleet)crates/harbor-cli— CLI binary (clap)crates/harbor-desktop— Tauri v2 desktop appui/— React frontend (Vite + Tailwind)
Build & Test
# Frontend
cd ui && npm ci && npm run build
# CLI
cargo build -p harbor-cli
# Desktop (requires Tauri prerequisites)
cd crates/harbor-desktop && cargo tauri dev
# Tests
cargo test --workspace
# Linting
cargo fmt --all -- --check
cargo clippy --workspace -- -D warnings
Key Conventions
- Nautical naming theme for CLI commands (dock, undock, fleet, launch, anchor, port, sync, lighthouse, scout, chest, cargo, crew)
- Config stored at
~/.harbor/config.toml - Connectors do safe merges into host configs (never overwrite)
- Syncing routes through the lighthouse gateway for tool filtering, vault resolution, and hot reload
- Host configs are auto-synced on dock/undock/toggle/connect
- Desktop app auto-starts the lighthouse on launch
- Users opt in to each host individually via
harbor port link <host>or the Link button in the UI vault:prefix in env vars references OS keychain secrets (resolved at runtime by the gateway)- Gateway starts HTTP server before initializing MCP servers (non-blocking)
Fleet Sync (harbor crew)
- Fleet repo lives at
~/.harbor/fleet/— a git repo containingharbor-fleet.toml - Fleet state (pull hashes) persisted at
~/.harbor/fleet-state.json harbor_core::fleetmodule:config,git,merge,provision,stateFleetServerDefis the shareable subset ofServerConfig— omitsenabled,auto_start,hosts,tool_hosts- Servers with
source = "fleet"in local config are fleet-managed - Hash-based drift detection: SHA-256 of
FleetServerDefstored after each pull; mismatches produceLocallyModifiedon the next pull (user changes are never silently overwritten) - Fleet git operations shell out to the
gitbinary (no libgit2) — leverages user's existing SSH keys and credential helpers - Desktop:
fleet_statusandfleet_pullTauri commands; server cards showfleet/modifiedbadges; Settings has a Crew section with Pull button
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.
- 5d ago First seen · 58 lines · 624 tokens per session scan A 02942572628f
Harbor CLAUDE.md is an instructions file published in the GitHub repository JoshuaShunk/Harbor (2 stars, last pushed 3mo ago), licensed MIT. It adds 624 tokens to every session, about $0.0031 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other instructions, from other repositories
skills-hub AGENTS.md
AGENTS.md instructions for qufei1993/skills-hub, covering skills hub - project rules, tech stack, common commands, directory structure and architecture.
skills-hub CLAUDE.md
Claude Code instructions for qufei1993/skills-hub: Read and follow all instructions in AGENTS.md.
agentbro AGENTS.md
Instructions for shirenchuang/agentbro, covering agents.md, what this is, local commands, mandatory github delivery workflow and before editing.
Ocul-PM CLAUDE.md
Claude Code instructions for bunhine0452/Ocul-PM, covering claude.md, what this is, commands, single frontend test and rust (run from src-tauri/).
cursor-studio copilot-instructions.md
Copilot instructions for DearLicy/cursor-studio, a project described as: Local-first Windows control center for Cursor: manage AI providers, balances, usage and costs, sessions, prompts, MCP, Skills, appearance, and integrations. | Cursor 本地桌面管理工具.
Ocul-PM AGENTS.md
AGENTS.md instructions for bunhine0452/Ocul-PM, covering ocul-pm 작업 기록 규칙, 0. 시작하기 전 — 과거를 먼저 찾는다, 1. 언제 기록하는가 (5 trigger), 2. 일지 파일 규격 (도구가 없을 때만) and 3. 금지.