woltspace CLAUDE.md

Development instructions for Woltspace, a platform that runs autonomous AI agents in isolated Docker environments and connects them to people through messaging apps.

In plain words
What is it for?
Use them when changing the Woltspace platform, its server, bot, command-line tool, Docker image, or the applications created by individual agents.
Why use it?
They give coding agents the project’s architecture, operating rules, and boundaries, including where their work may be made.

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/jerpint/woltspace/claude-md
Clone the repo
git clone --depth 1 https://github.com/jerpint/woltspace
Per session 5,476 This file is loaded in full into every session.
When invoked 5,476 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 $0.05476 $0.05476
Opus 5 $0.02738 $0.02738
Sonnet 5 $0.01095 $0.01095
Haiku 4.5 $0.00548 $0.00548

Measured 3d ago against content hash aeccd231e3d7, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

woltspace 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 3d 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 · 421 lines

How it starts

The opening of the file, as written. The whole thing — 421 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Woltspace — Developer Guide

What is this?

Woltspace is a platform for running autonomous AI agents called wolts. Each wolt is:

  • An AI with a persistent identity, memory, and personality
  • Living inside a Docker container (the "space") where it can build whatever it wants
  • Connected to its human partner via messaging apps (Telegram, WhatsApp/Slack)

The human chats with the wolt via Telegram. The wolt can spin up Claude Code sessions to do real work — write code, build websites, generate content. The human can watch the work happen live via a split-view browser UI (terminal on left, viewport on right).

This repo is the platform — the Docker image, server, bot brain, and CLI that makes wolts run. Individual wolt data lives separately (under ~/.woltspace/wolts/{name}/).

⚠️ Platform code is immutable from wolt sessions. Wolts must NEVER edit files in /workspace/woltspace/. All wolt work happens inside their own directory (/workspace/wolts/{name}/). Apps go in wolts/apps/. If a wolt needs new platform functionality, the human files an issue — wolts don't patch the platform.


Architecture at a Glance

Human (Telegram/Slack)
    ↓
Python bot (litellm → Claude/Haiku)
    ↓
Tool: claude_code → tmux session → Claude Code CLI
    ↓
Wolt builds things (site, sparks, artifacts)
    ↓
FastAPI server (port 7777) → split view in browser ← Human watches
    ↑
cloudflared tunnel (optional public URL, no account needed)

Key design: The woltspace repo is git clone'd into the Docker image at build time. The only host mount is ~/.woltspace/wolts/ — all wolt data. The container is disposable; the wolts directory is the backup.


Lore — The Colony

Woltspace is the whole thing. Wolts live and work here.

The lodge — home base. Always-on, handles day-to-day chat.

The dens — where real work happens. Claude Code sessions, spawned per task, dissolved when done.

The pond — the visible surface (viewport). Wolts build in dens, surface work to the pond.

Read the full file on GitHub · 421 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. 3d ago First seen · 421 lines · 5,476 tokens per session scan A aeccd231e3d7

Subscribe to this mod's changes

woltspace CLAUDE.md is an instructions file published in the GitHub repository jerpint/woltspace (11 stars, last pushed 3d ago), licensed MIT. It adds 5,476 tokens to every session, about $0.0274 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 2026-08-30.