Hive CLAUDE.md

Hive CLAUDE.md is an instructions file for coding agents from chiruu12/Hive. It costs 2,312 tokens per session, scanned A, original, MIT.

Project instructions for Hive, a local-first system for running persistent AI agents that collaborate, write code, and use tools. They describe its architecture, technologies, and coding rules.

In plain words
What is it for?
Understanding Hive's folders, configuring development, and following its code and architecture guidelines.
Why use it?
They give contributors the project context and conventions needed to make consistent changes.

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/chiruu12/hive/claude-md
Clone the repo
git clone --depth 1 https://github.com/chiruu12/Hive

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 Hive CLAUDE.md

README.md
[![agentmods](https://agentmods.dev/badge/instructions/chiruu12/hive/claude-md.svg)](https://agentmods.dev/instructions/chiruu12/hive/claude-md)
Your own site
<a href="https://agentmods.dev/instructions/chiruu12/hive/claude-md"><img src="https://agentmods.dev/badge/instructions/chiruu12/hive/claude-md.svg" alt="Measured on agentmods" height="20"></a>
Per session 2,312 This file is loaded in full into every session.
When invoked 2,312 The same file — it is already loaded in full.
Security scan A 0 findings. 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 $0.02312 $0.02312
Opus 5 $0.01156 $0.01156
Sonnet 5 $0.00462 $0.00462
Haiku 4.5 $0.00231 $0.00231

Measured 5d ago against content hash 83bfb6de2a3a, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

Hive 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.

CLAUDE.md · 186 lines

How it starts

The opening of the file, as written. The whole thing — 186 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Hive Development

Project Overview

Hive is a local-first agent operating system. Users spawn persistent AI agents via CLI that collaborate, write code, and use tools autonomously. Config-driven with YAML agent profiles, multi-model support (Anthropic, OpenAI, Fireworks, Ollama, LM Studio), and built-in safety via workspace isolation.

Architecture

src/hive/
├── cli/              # Typer CLI (commands map 1:1 to user actions)
├── daemon/           # Background service (lifecycle, heartbeat loop, diagnostics)
├── agents/           # Agent profiles, state, suffering, identity, delegation
├── runtime/          # Standalone agent framework (ReAct loop, tools)
├── memory/           # Persistence (SQLite store, JSONL events, semantic memory)
├── models/           # Model providers (Anthropic, OpenAI, Groq, etc.), registry, factory
├── interactions/     # Multi-agent interaction patterns (round-table, pairs, freeform)
├── world/            # Simulated economy (jobs, skills, finances, life events)
├── logging/          # Structured run logs (decisions, tools, goals, suffering)
├── mcp/              # MCP client and server
├── server/           # REST API (FastAPI) -- optional [api] extra (hive serve)
├── checkpoint.py     # Save/restore agent state snapshots
├── api.py            # Programmatic Python API (Hive facade class)
└── config.py         # Config loading (YAML + env vars)

Tech Stack

  • Language: Python 3.11+
  • CLI: Typer + Rich (terminal UI)
  • Async: asyncio (concurrent agent execution)
  • Database: SQLite via aiosqlite
  • LLM: anthropic SDK, OpenAI SDK (also for Fireworks, Ollama, LM Studio)
  • Packaging: uv, pyproject.toml
  • Testing: pytest + pytest-asyncio

Code Conventions

Architecture Principles

  • Config-driven: Agent behavior defined in YAML, not hardcoded
  • OOP for entities: Agent, Task, Room, Tool are classes with clear interfaces
  • FP for data transforms: Pure functions for parsing, routing, formatting
  • Protocol classes: Use Python Protocols for model providers, tool executors
  • Dependency injection: Components receive dependencies, don't create them
  • Single responsibility: Each module does one thing well

Read the full file on GitHub · 186 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. 5d ago First seen · 186 lines · 2,312 tokens per session scan A 83bfb6de2a3a

Subscribe to this mod's changes

Hive CLAUDE.md is an instructions file published in the GitHub repository chiruu12/Hive (5 stars, last pushed 12d ago), licensed MIT. It adds 2,312 tokens to every session, about $0.0116 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.