Start Here
Operator Journey
A single recommended path for new operators: install, configure, validate, and run safely.
- Install + bootstrap: run
wrenos initand generate baseline config. - Safety checks: run
wrenos doctorand resolve all red diagnostics. - Runtime verification: run
wrenos statusand confirm paper-first execution posture. - Configuration hardening: set required env/config values and re-run checks.
- Launch readiness: validate against the release checklist and regression artifacts before production tags.
Quick Start → Configuration → Safety Posture → CLI Overview → Architecture.
WrenOS Documentation
WrenOS Documentation
The open-source control plane for operator-managed crypto agent systems.
/system inventory
- packages6
- core_skills_pack13 audited
- default_mcp_servers3 wired by init
- agent_packs2 active
Explore the Docs
Getting Started
Install fast, initialize a profile, and validate runtime health.
CLI Reference
Command model, init flow, diagnostics, and config controls.
Packages
Understand core modules, profiles, loops, and adapters.
Private Inference (Optional)
Optional private inference route; default v1 path does not require Speakeasy.
Architecture
Agent coordination, tool layers, and execution boundaries.
Safety & Risk
Paper-first controls, approvals, and custody separation.
Docs Parity Matrix
Section-by-section map of live docs vs source markdown paths.
Docs content synced against wrensignal/wrenOS@a73300c.
Site Ownership Map
Source-of-truth routing for wrenos.ai surfaces and repo ownership.
Getting Started
Quick Start
Install WrenOS, initialize an agent profile, and verify your local stack in minutes.
Prerequisites
- Node.js 20+ and npm
- OpenClaw runtime access
- A Solana wallet for eventual live execution (paper mode is default)
Install and initialize
git clone https://github.com/wrensignal/wrenOS.git
cd wrenOS
npm install
node packages/cli/src/index.mjs init --profile research-agent
node packages/cli/src/index.mjs doctor
node packages/cli/src/index.mjs status
What init does
Creates project config, wires defaults, and scaffolds MCP definitions for research, tool access, and execution context. If .mcp.json is missing, it generates one.
Next steps
Keep paper mode enabled until your strategy passes repeated backtests and paper sessions under varied conditions.
| Goal | Command |
|---|---|
| Inspect config | node packages/cli/src/index.mjs config list |
| Check inference link | node packages/cli/src/index.mjs test inference |
| Check execution path | node packages/cli/src/index.mjs test execution |
Getting Started
Safety Posture
Operational safeguards for agent-managed strategy systems.
Paper-first by default
WrenOS starts with simulation mode enabled. The system records strategy decisions and execution intent without committing capital, giving operators a full audit trail before live risk.
Approval model
External side effects can require explicit approvals based on policy. This keeps execution deliberate and reviewable as confidence changes.
Do not enable live execution until you validate tool integrity, routing stability, and strategy behavior across adverse market conditions.
Custody separation
WrenOS does not require custody of private keys. Signing remains in your wallet boundary while agents handle analysis, scoring, and execution decisions.
Degrade safely
Confidence tiers can tighten or suspend execution as data quality deteriorates, reducing impulsive behavior in uncertain states.
Edit this page on GitHubCLI
CLI Overview
Core command surface for setup, diagnostics, runtime checks, and configuration.
Command model
The CLI is designed around initialization, verification, and controlled iteration. Commands are composable and readable, with explicit naming for operator clarity.
Common commands
# initialize a profile
wrenos init --profile research-agent
# run health checks
wrenos doctor
wrenos status
# inspect/set config
wrenos config list
wrenos config set liveExecution false
wrenos init
Bootstraps project configuration and writes default runtime artifacts.
wrenos doctor
Runs diagnostics for dependency health, connectivity, and required runtime components.
wrenos status
Returns current profile, routing state, and execution posture.
wrenos config
Reads and updates config keys for inference, execution, and policy controls.
wrenos start
Starts the configured runtime loop for your selected profile.
Edit this page on GitHubGetting Started
Configuration
WrenOS uses inspectable JSON config under .wrenos/config.json (legacy .0xclaw/config.json fallback remains temporary).
Config priority
Runtime reads .wrenos/config.json first. If missing, it can read the legacy path and emits a migration warning. For fresh setups, always use .wrenos/.
Common keys
| Key | Purpose |
|---|---|
profile | Selected operator profile template. |
liveExecution | Global safety gate. Defaults false. |
inference.baseUrl | Inference routing endpoint (default open route; private route optional). |
inference.routes.* | Route families for research/deep_think/codegen/uncensored. |
loop.* | Cadence + execution behavior. |
execution.venues.* | Execution venue adapters and toggles. |
Set values from CLI
wrenos config set liveExecution false
wrenos config set inference.baseUrl https://your-openai-compatible-endpoint.example
wrenos config set execution.venues.jupiter.enabled trueCLI
wrenos init
Initialize WrenOS with a profile and write safe defaults.
Usage
wrenos init --profile research-agentWhat it does
- Creates
.wrenos/config.jsonfrom profile templates. - Sets
liveExecution: falseby default. - Generates
.mcp.jsonif missing.
CLI
wrenos doctor
Run environment and configuration diagnostics.
Checks include
- Node runtime compatibility
- Config presence/profile validity
- MCP config availability
- Inference route readiness
- Execution mode + paper-safe default
Use wrenos doctor before every profile switch or deploy.
CLI
wrenos status
Snapshot current profile, safety posture, and runtime-facing config.
Purpose
Use status to confirm selected profile, execution mode, and critical route values before running loops or enabling any external action path.
Edit this page on GitHubCLI
wrenos config
Read and mutate nested config keys using dot-path syntax.
Usage
wrenos config set risk.maxTradeUsd 25
wrenos config set loop.heartbeatAdaptive true
wrenos config set inference.routes.research deepseek-v3.2CLI
wrenos start
Beta orchestration loop for structured heartbeat operations.
Current status
wrenos start is available but still beta. Prefer deterministic checks with wrenos start --once and supervise long-running production loops with your scheduler/process manager.
Packages
Core
@wrenos/core — policy defaults, safety primitives, and shared utilities.
Role
Core centralizes reusable control-plane behaviors and safety defaults so profile and adapter layers share consistent policy semantics.
Edit this page on GitHubPackages
Profiles
@wrenos/profiles — paper-first, safety-first profile templates.
Shipped templates
- research-agent
- research-only
- solo-trader-paper
- trading-agent-paper
- trading-agent-live-disabled
- meme-discovery-research
- meme-discovery-trading-paper
Packages
Loops
@wrenos/loops — heartbeat planning and validation summaries.
Role
Loop primitives govern cadence and condition-aware behavior, helping operators degrade safely as data quality or route health changes.
Edit this page on GitHubPackages
Adapters
@wrenos/adapters — connectors for inference, execution, and operator interfaces.
Adapter scope
- Inference: OpenAI-compatible request shape (private route optional)
- Execution: Jupiter + pump.fun paths
- Operator UX: chat/operator surfaces
Packages
Private Inference (Optional)
OpenAI-compatible private-route SDK support (Speakeasy optional, not required for v1 baseline).
Install
npm install speakeasy-aiWhat it handles
- Initial request
- 402 challenge parsing
- EIP-712/EIP-3009 signing
- Replay with payment proof headers
- JSON/SSE streaming output
Packs
Meme Discovery
Dual-agent meme discovery workflow with composite scoring.
Init
wrenos init-pack --pack meme-discoveryIncludes
- Research + trading agent configs
- Watchlist handoff schema (
inbox/meme-watchlist-*.json) - Composite weighted scoring defaults
- Jupiter + pump.fun adapter wiring
Packs
Research Agent (Dual Agent Pack)
Bare-bones two-agent operating model for research-to-trading handoff.
Components
- research-agent for evidence briefs
- trading-agent for validation and risk-gated execution posture
- handoff schema at
handoff/research-brief.schema.json
Guides
Private Inference (Optional)
Optional private inference routing with explicit fallback posture.
Baseline env
OPENAI_BASE_URL(or default route)SPEAKEASY_API_KEY(deployment-dependent)- route envs/overrides as needed
Operational guardrails
- Cache deterministic calls where possible
- Gate expensive calls on material input deltas
- Track per-cycle model usage and cost
- Report degraded route state explicitly
Guides
Railway Deploy
Template deploy for a full WrenOS stack with paper-first defaults.
Required env
PROFILEOPENAI_BASE_URL(or default route)OPENCLAW_API_KEYTELEGRAM_BOT_TOKEN(for chat UX)
Post-deploy verification
- Service health and restart stability
- Inference connectivity and route checks
- Profile + execution mode sanity
- Heartbeat/log artifact verification
Guides
Migrating from 0xClaw
Move to WrenOS naming and config paths with minimal disruption.
Key changes
- Repo:
wrensignal/0xClaw→wrensignal/wrenOS - CLI:
0xclaw→wrenos - Config:
.0xclaw/→.wrenos/
Recommended migration
wrenos migrate
wrenos doctor && wrenos statusGuides
Docs Parity Matrix
Section-by-section parity map between wrenos.ai/docs and source markdown in wrensignal/wrenOS.
| Live section | Source of truth | Status |
|---|---|---|
| Quick Start | docs/quickstart.md | Aligned |
| Safety Posture | docs/safety.md | Aligned |
| CLI acceptance | docs/cli-acceptance.md | Referenced |
| MCP diagnostics | docs/mcp-doctor-signatures.md | Referenced |
| Privy onboarding | docs/privy-wallet-provisioning.md | Referenced |
| Railway deploy | docs/railway-first-run-playbook.md + RAILWAY_DEPLOY.md | Aligned |
| Profile matrix | docs/profile-matrix.md | Aligned |
| Architecture/runtime | docs/architecture.md, docs/runtime-status.md | Aligned |
| Migration guides | docs/migrating-from-0xclaw-to-wrenos.md | Aligned |
Reviewer checklist
- Every major docs section links to a concrete source markdown path.
- No claim contradicts current CLI/runtime behavior.
- Command snippets match current command names and flags.
- Safety defaults remain paper-first and opt-in for live execution.
Reference
Architecture
Composable control plane: profiles, adapters, loops, and operator governance.
Three-layer model
- Open-source control plane (CLI, profiles, packs, loops, adapters)
- Hosted runtime infrastructure (optional/default path)
- Operator workflows and governance (approvals, runbooks, audits)
Runtime flow
Inference path: agent loop → adapter → default inference route (optional private route supported).
Execution path: decision → execution adapter → venue quote/build → paper/live policy gate.
Edit this page on GitHubReference
Runtime Status
Current maturity snapshot from the repository docs.
Usable today
- CLI lifecycle: init, doctor, status, config, wallet, test, start, init-pack, bootstrap-wrenos
- Profile templates with paper-mode defaults
- Inference connectivity tests and execution path tests
- Inspectable file-based configuration and output artifacts
Experimental
wrenos startbeta orchestration loop- Turnkey Telegram operator stack as a production-default path
- Expanded pack ecosystems and auto-wiring assumptions
Reference
Config Schema
Representative schema derived from shipped profile templates.
Example profile JSON
{
"profile": "trading-agent-paper",
"liveExecution": false,
"loop": { "enabled": true, "heartbeatAdaptive": true, "tradeExecution": "paper" },
"risk": { "maxTradeUsd": 25, "maxDailyNotionalUsd": 75 },
"inference": {
"provider": "openai-compatible",
"baseUrl": "https://your-openai-compatible-endpoint.example",
"routes": {
"research": "deepseek-v3.2",
"deep_think": "qwen3-235b-a22b-thinking-2507",
"codegen": "qwen3-coder-480b-a35b-instruct",
"uncensored": "venice-uncensored"
}
}
}Reference
Changelog
Recent release history.
2026-03-07 — Rebrand: 0xClaw → WrenOS
- Primary CLI renamed to
wrenos - Primary config directory renamed to
.wrenos/ - Workspace package scope moved to
@wrenos/*
Compatibility aliases remain temporarily (0xclaw, legacy config fallback, bootstrap aliases), with planned removal target v0.3.0.
WrenOS