Skip to content
Reyn
Reyn
Stdlib
English
日本語
Initializing search
Architecture
Docs
GitHub
Home
Start here
Feature Map
Guide
Getting started
Installation & setup
01 — Installation
02 — Chat mode
For users
Using Reyn
Chat and Web UI
How to rewind and branch
Manage permissions
Configure the sandbox
Cap your spending
Files & integrations
Work with local files
Enable semantic action search
Popular local MCP servers — install + usage procedure
Log in to an OAuth provider
Inspect and manage memory
For Reyn developers
Concepts & principles
P1–P8 and the code that enforces them
Extending Reyn
Add a new Control IR op kind
Write deterministic tests with LLMReplay
Benchmarking
Run SWE-bench
Cookbook
Reyn cookbook
Examples
research-topic
translate-doc
weekly-summary
write-readme
Configs
Reference
CLI
Running Reyn
reyn run-once
reyn chat
reyn dogfood
reyn web
Multi-agent & tools
reyn agent
reyn topology
reyn mcp
reyn embeddings
Configuration & auth
reyn config
reyn permissions
reyn secret
reyn auth
reyn init
Session & ops
reyn events
reyn memory
reyn cron
reyn source
Common flags
Runtime
Context frame
LLM output contract
Control IR
Events
Config
Core
reyn.yaml
Permissions
Budget and cost tracking
Infrastructure
.reyn/ — state directory
Multi-agent config
Built-in model catalog
Stdlib
Testing
reyn.dev.testing.LLMReplay API
Test Tier Auditor (scripts/test_tier_audit.py)
Replay & Tracing
LLM Payload Tracing (Dogfood Debug)
SP Render Inspector (scripts/dogfood_sp_render.py)
Contributor Tools
Agent Worktree Cleanup (scripts/cleanup_agent_worktrees.py)
Other
Glossary — canonical reyn terms
Upgrade policy
Concepts
Architecture
Care boundary — what Reyn cares about, what it doesn't
LLM invocation surfaces — Router-style vs Phase-style
Agent interaction layers
LLM as a constrained decision engine
Sandbox and permissions: orthogonal concerns
Runtime
Workspace
Events
Agent lifecycle hooks
Config hot-reload
Time-Travel: Rewind and Resume
Crash-recovery completeness and the durability boundary
Permission model
Capability profile
Delegation policy
Sandbox
Secret handling
Multi-agent
Sessions and the agent–session model
Multi-agent
A2A (Agent2Agent Protocol)
Topology
LLM org-design tools
Data & retrieval
RAG (Retrieval-Augmented Generation)
Memory
Operational Intelligence
Chat compaction
Tools & integrations
MCP (Model Context Protocol)
Skills
Universal Action Catalog
Tool-Use Schemes
CodeAct: Code-as-Tools
Voice input
Observability
Dogfood Scenario Framework
Agent engineering — six lenses
System Design
Tool Contract Design
Retrieval Engineering
Reliability Engineering
Security
Product Think
Stdlib
¶
Bundled components shipped with reyn.
On this page
Stdlib
Back to top