Signal Rail is a document governance system for personal and AI-assisted projects. Instead of letting notes, decisions, blockers, and ideas pile up in one undifferentiated space, Signal Rail gives each kind of material its own rail — a dedicated plain-text file with a defined role, routing rules, and authority boundary. The result is a project that stays readable as it grows, branches, and comes back after interruption.Documentation Index
Fetch the complete documentation index at: https://mintlify.com/xxyoudeadpunkxx/signal-rail/llms.txt
Use this file to discover all available pages before exploring further.
Introduction
Understand what Signal Rail is, why it exists, and how the rails work together.
Quickstart
Deploy a Signal Rail instance and open your first AI-assisted working session in minutes.
The Rails
Reference for every canonical file: its role, what belongs there, and what does not.
Operating with AI
How the lateral kernel governs agent behavior, modes, routing, and write gates.
How Signal Rail works
Every piece of project material has a nature — is it current live work, a decision already taken, a hard identity constant, or a still-open idea? Signal Rail asks that question first and routes the material to the right file before it can become noise.Deploy an instance
Run
init_signal_rail.bat (Windows) or copy the canonical files manually into your project folder. The bootstrap script creates all thirteen files and pre-fills the date and project name.Open with an AI agent
Ask your AI assistant to read
00_runtime_entry.txt, then 06_ai_to_ai.txt. The agent closes the minimum working frame — host project, working object, mode, and source authority — before taking any action.Fill the two live rails first
Write
01_orientation.txt to describe what the project is. Write 03_master_working.txt to capture the current live state, dominant blocker, and next sensible move. These two files are the minimum read for every session.The rails at a glance
| Rail | File | What belongs there |
|---|---|---|
| 🚪 Runtime entry | 00_runtime_entry.txt | Entry validity, minimum read order, permissions |
| 🧭 Orientation | 01_orientation.txt | Project identity, perimeter, reading frame |
| 🧊 Freeze | 02_protocol_freeze.txt | Hard-to-reopen identity constants (max 5) |
| 🔥 Master working | 03_master_working.txt | Current live state, blocker, next step |
| ⚖️ Decision log | 04_decision_log.txt | Already-won decisions in effect |
| 🌱 Latent ideas | 05_latent_ideas.txt | Live unresolved material |
| 🧠 Lateral kernel | 06_ai_to_ai.txt | AI operating behavior inside the instance |
| 🧰 Guided prompts | 07_guided_prompts_test.txt | Safe start, rebuild, review, and routing flows |
| 🗺️ Surface map | 08_surface_map.txt | Technical topology, entrypoints, minimal runbook |
| 🔁 Handoff | 09_handoff_reentry.txt | Session continuity and re-entry support |
| 🧲 Field findings | 97_field_findings.txt | Lateral captures before routing or discard |
| 📌 Parking | 98_parking.txt | Useful but not active material |
| 📦 Archive | 99_archive.txt | Closed, historical, or no-longer-live material |
Tools & Bootstrap
Windows batch and PowerShell scripts for deploying new Signal Rail instances.
Offline Workstation
The single-file HTML workstation for reading, writing, and managing a live instance.
Guided Prompts
Structured prompt flows for safe session starts, confused sources, and retro timelines.
Core Concepts
Routing discipline, authority hierarchy, canonical levels, and the baseline vs. instance model.