Documentation

Sonderr Documentation

Everything you need to install, master and trust Sonderr — the privacy-first AI workspace. Twenty-eight deep guides, organized the way you’ll actually use them: get running, learn the workflow, unlock the capabilities, verify the trust. Long enough to get lost in.

GETTING STARTED 28 GUIDES OPEN SOURCE About Sonderr

Quick start

curl -fsSL https://raw.githubusercontent.com/dxn111/sonderr/main/install.sh | sh
sonderr          # starts the local runtime on http://127.0.0.1:4173

Requires Node.js 20+, npm and Git. The browser workspace opens automatically — that tab is the product.

Get started

From zero to a running workspace in one command — then everything about how it’s wired.

Install & first run
START HERE

Requirements, the one-command installer, PATH setup, launch flags, updates and development checkouts.

Quickstart — your first 30 minutes
FAST

Install, launch, free model with no API key, the four modes, your first real task, and where everything lives.

CLI & launcher reference
REFERENCE

The launcher philosophy, install-script internals, --no-open and --port, the fail-closed update gate and the full data map.

Configuration & environment
SETUP

Launch flags, file locations, required updates, email env vars, wallet profiles and development checkouts.

The workflow

How Sonderr actually works: the four task modes, memory that survives days, and the surfaces where real work happens.

Web workspace handbook
CORE

Task modes, composer shortcuts, artifacts, uploads, task lists, Studios, Plugin Hub and collaboration rooms.

Prompting & task briefs
WORKFLOW

Pick the mode before the words, the four-part brief pattern, weak briefs made strong, and why constraints become acceptance checks.

Artifacts, files & deliverables
OUTPUT

Download cards, the artifact panel, uploads, and where every deliverable actually lives on your machine — the full path in and out.

Sessions, checkpoints & long tasks
MEMORY

Persistent sessions, live task lists, compaction, checkpoints, resume after crashes and honest work budgets.

Studios — goal to deliverable
RC

Seven studios, two-specialist crews, per-craft playbooks with a definition of done, milestone boards with checkpoints.

Agent teams & collaboration rooms
SWARM 2.0

One lead, six read-only workers, findings and risk flags, the published synthesis pass and the honesty contract.

The four modes, deeply
PERMISSION

Ask, Plan, Build and Vision — what each mode is allowed to cause, when to reach for it, and the mistakes that waste turns.

Marathon tasks — checkpoints & resume
LONG-RUN

Durable checkpoints, honest compaction, 72-hour marathon tasks and the resume offer after a crash. Work that outlives a context window.

Capabilities

Everything the agent can do once it’s running — models, images, skills, the web, and the doors to your stack.

Models & providers
AI

Kilo Gateway and free-tagged models, API keys, the local Sonderr-v1 model and what leaves your machine.

Image generation & Vision mode
CREATE

The edit_image tool, three permission gates, OpenAI-compatible image endpoints, download-card delivery and pinned web fetches.

Skills — 65 playbooks
LIBRARY

Markdown playbooks for build, craft, data, ops and safety work — matched to your task and loaded at the moment they matter.

Web research & live sources
TOOLS

Bounded, read-only web access built in — verified-domain filters, blocked private hosts, no MCP setup.

MCP — Model Context Protocol
EXTEND

Local and remote MCP servers, stdio and HTTP transports, DNS-pinned endpoints and present-turn authorization for mutations.

Plugins, connectors & email
CONNECT

The Plugin Hub, six connectors — Gmail, Notion, Slack, Drive, GitHub, Linear — and send-side email that drafts in the open.

Wallets & the Web3 toolset
OPT-IN

EVM and Solana, prepare-then-confirm actions, keys that never render in chat and the hard lines that never move.

Recipes & reference

Whole workflows you can steal, and the vocabulary that keeps the rest honest.

Recipes — eight real-world workflows
RECIPES

Pre-change investigation, pre-ship review rooms, Studios websites, print-ready documents, research briefs, data cleanup, image work, the weekly sweep.

The Sonderr glossary
REFERENCE

Every term the product uses, defined honestly — including the ones that sound grander than they are.

Planning a project end to end
BLUEPRINT

Brief, plan, milestones, checkpoints, review and a definition of done re-run as the final acceptance check. Steal the shape.

Documents, reports & print-ready output
WRITING

Artifacts you can preview and download, the print-ready pipeline, and reports that survive review — with receipts.

Research that holds up
RECEIPTS

Bounded, read-only web access, a six-worker research crew and a synthesis pass that reports gaps instead of hiding them.

Trust & operations

What happens on your machine, what never leaves it, what to do when something breaks — and the rhythm every release answers to: slow by design.

Security, updates & privacy
TRUST

Local-first architecture, secret scrubbing, workspace sandboxing, the required-update gate and wallet safety.

Your data, your machine
LOCAL-FIRST

The complete data map, what crosses the wire and what never does, backup in one copy, moving machines, and how to wipe it all.

Troubleshooting & support
HELP

Port conflicts, PATH issues, update-gate recovery, model availability and how to reach a human fast.

Official sources
REPO

README, CHANGELOG, install.sh and the full source — Sonderr ships in the open under MIT.

Updates, versions & the rhythm
RHYTHM

The fail-closed update gate, required updates, what a release touches — and the honest shipping rhythm behind them: slow by design.

Studio & policies

What we do — and don't · Privacy (we collect nothing) · Terms · Acceptable Use · Licensing · Responsible Disclosure · Security · Imprint

Need a human? Email dxn1.0day@gmail.com or danipostma123@gmail.com. Security issues skip the queue — see Responsible Disclosure.