1.4 AI-Assisted Play

Updated: v2026.02.02

Aurora’s depth creates a paradox: the comprehensive documentation needed to play well can itself become overwhelming. This section describes how to use AI assistants (Claude, ChatGPT, or similar) as a real-time coach, leveraging this manual as a verified knowledge base.

Contents

Updated: v2026.02.02

1.4.1 Why AI-Assisted Play Works for Aurora

Updated: v2026.02.02

Aurora is uniquely suited to AI assistance because:

  • Turn-based pacing: You can pause indefinitely to consult your AI coach without time pressure
  • Database-driven mechanics: Most game rules are deterministic formulas that can be precisely explained
  • Dense documentation: This manual exists specifically to provide AI with verified, citation-backed information
  • Complex interdependencies: AI excels at explaining how system A affects system B affects system C

AI assistance does not replace learning the game — it accelerates it by providing context-aware guidance when you need it.

1.4.2 Quick Start: The /coach Skill

Updated: v2026.02.01

If you’re using Claude Code with this manual repository, the /coach skill provides the fastest path to AI-assisted play. It automatically queries your game database and provides spoiler-free coaching.

Usage 

/coach [optional-database-path]

If no path is provided, the coach looks for the database at ../Aurora/AuroraDB.db relative to the manual directory.

Example:

/coach ~/Downloads/Aurora271Full/AuroraDB.db

What /coach Does

  1. Queries your game database — Automatically reads your AuroraDB.db to understand your current situation
  2. Detects your game phase — Identifies whether you’re in conventional start, early TN, exploration, expansion, or contact preparation
  3. Provides phase-appropriate guidance — Recommends priorities based on where you are in the game
  4. References this manual — Cites specific sections for detailed mechanics explanations
  5. Maintains spoiler discipline — Never reveals NPR information, unexplored systems, or hidden events

Game Phase Detection

The coach automatically determines your game phase:

Phase Indicators Coach Focus
Phase 0: Conventional No TN tech researched Research priorities, CI conversion
Phase 1: Early TN TN tech done, no ships Ship design, shipyard retooling
Phase 2: Exploration Survey ships built Survey mechanics, fuel harvesting
Phase 3: Expansion Jump-capable Colony setup, mining operations
Phase 4: Contact Prep Multiple systems Combat mechanics, sensor coverage

Spoiler Protection

The coach enforces strict information boundaries:

Will Query:

  • Your race’s colonies, ships, and fleets (RaceID with NPR=0)
  • Your completed research and active projects
  • Your shipyards, installations, and mineral stockpiles
  • Your commanders and their assignments

Will NOT Reveal:

  • NPR locations, fleet compositions, or tech levels
  • Contents of unexplored systems
  • Jump point destinations you haven’t surveyed
  • Precursor, Invader, Swarm, or Eldar information
  • Hidden event outcomes

Response Format

Coach responses follow a consistent structure:

  1. Current Situation — Brief summary of your game state
  2. Immediate Priority — The single most important thing to do next
  3. Next Steps — 2-3 follow-up actions after the priority
  4. Learn More — Links to relevant manual sections

When to Use /coach

  • Starting a new session — Get oriented on what you were doing and what to prioritize
  • Feeling overwhelmed — Let the coach identify your single most important next action
  • Transitioning phases — Understanding what changes when you finish TN research, build your first ships, or make first contact
  • Learning the game — The coach explains mechanics while pointing you to detailed manual sections

For urgent tactical situations (NPR invasion, combat in progress), the coach provides guidance but cannot make real-time decisions — you’ll need to apply the advice yourself.

Note: If you don’t have Claude Code, see Section 1.4.3 for manual setup instructions.

1.4.3 Manual Setup

Updated: v2026.02.01

If you’re not using the /coach skill, you can still get excellent AI assistance with manual setup.

1.4.3.1 Providing Context

For best results, give the AI access to this manual. Options include:

  • Direct file access: Tools like Claude Code can read the manual files directly
  • Repository reference: Point the AI to the GitHub repository (github.com/ErikEvenson/aurora-manual)
  • Copy relevant sections: Paste specific manual sections into your conversation when asking about those topics

1.4.3.2 Establishing Game State

Begin each session by telling the AI your current situation:

I'm playing Aurora C# v2.7.1 with a conventional start.
Current year: 2031
Population: 850 million on Earth
I have 400 construction factories and 200 mines.
I just researched my first engine technology.

The more context you provide, the more relevant the AI’s guidance will be.

1.4.3.3 Direct Database Access

AI assistants with file system access (such as Claude Code) can query your game database directly. The Aurora database (AuroraDB.db) is a SQLite file containing all game mechanics, your current game state, and reference data.

Location: Typically in your Aurora installation folder (e.g., ~/Downloads/Aurora271Full/AuroraDB.db)

What AI can query directly:

  • Reference tables (DIM_*): Component stats, tech trees, installation costs, mineral requirements
  • Game state (FCT_*): Your ships, colonies, populations, fleets, research progress
  • Tech system (FCT_TechSystem): All technologies with costs, prerequisites, and effects

Example prompts using database access:

Query my game database and tell me:
- How many construction factories do I have on Earth?
- What technologies have I researched?
- What ships are in my fleet?

Look up the exact mineral cost for a Fuel Refinery
in DIM_PlanetaryInstallation and tell me what I need.

Advantages of direct database access:

  • Real-time accuracy: AI sees your actual game state, not your description of it
  • No transcription errors: Values come directly from the database
  • Deep queries: AI can correlate data across tables (e.g., “which of my ships are low on fuel?”)
  • Verification: AI can confirm manual claims against the actual game database

Setup for Claude Code:

My Aurora database is at: ~/Downloads/Aurora271Full/AuroraDB.db
You can query it with sqlite3. Use this for accurate game data.

Example database query and output:

When you ask “What are the mineral costs for a Fuel Refinery?”, an AI with database access can run:

sqlite3 AuroraDB.db "SELECT Name, Cost, Duranium, Neutronium, Boronide
FROM DIM_PlanetaryInstallation WHERE Name='Fuel Refinery';"

Output:

Fuel Refinery|120.0|0.0|0.0|120.0

This shows directly from the database: 120 BP cost, 120 Boronide required, no Duranium or Neutronium. No ambiguity, no outdated wiki information.

1.4.4 Prompting Patterns

Updated: v2026.02.02

Note: The /coach skill (Section 1.4.2) automates situation assessment and priority recommendations. Use these prompting patterns when you need specific answers beyond what the coach provides, or when not using Claude Code.

Figure 1.4.4: AI Coaching Workflow

The workflow above illustrates the decision process for AI-assisted play. The key insight: always verify AI responses against citations before committing to major decisions, but allow quick action in urgent situations.

1.4.4.1 Situation Assessment

When you’re unsure what to focus on:

Given my game state [describe], what should my priorities be
for the next 5 years? Reference the manual for recommended
build orders or strategic guidance.

1.4.4.2 Problem Diagnosis

When something seems wrong:

My ships keep running out of fuel before reaching their
destination. What am I doing wrong? Cite relevant manual
sections about fuel consumption and ship design.

1.4.4.3 Decision Support

When facing a choice:

I can build either 50 more mines or 20 more construction
factories. Which is better for my current situation?
Explain the trade-offs using the manual's economy section.

1.4.4.4 Mechanic Explanation

When you need to understand a system:

Explain how maintenance facilities and [MSP](/aurora-manual/appendices/B-glossary.html#msp) (Maintenance Supply Points) production work
in C# Aurora. I keep seeing "lack of MSP" warnings.
Cite the manual sections and any relevant database values.

1.4.4.5 UI Navigation

When you can’t find something:

I need to design a ground force formation. Which window
do I open and what buttons do I click? Walk me through
the process step by step.

1.4.5 Session Workflow

Updated: v2026.02.01

1.4.5.1 Pre-Session Briefing

Before playing:

I'm about to start a new Aurora session. Last time I:
- Colonized my first moon
- Started building a shipyard
- Detected an NPR 3 jumps away

What should I focus on this session? What threats should
I be watching for?

1.4.5.2 Mid-Session Consultation

During play, keep the AI conversation open. Ask quick questions as situations arise:

  • “An NPR fleet just appeared in my home system. What do I do?”
  • “My colony is showing 40% unrest. How do I fix this?”
  • “I need to design a missile. What are the key trade-offs?”

1.4.5.3 Post-Session Review

After playing:

I just finished a session where [describe what happened].
What did I do well? What mistakes did I make? What should
I read in the manual before my next session?

1.4.6 Example Sessions

Updated: v2026.02.01

1.4.6.1 Early Game: First TN Research

The following illustrates a typical AI-assisted play session:

Player: I’m in year 2028, conventional start. I just finished researching Trans-Newtonian Technology. What do I do now?

AI: Congratulations — this unlocks the entire TN economy. Your immediate priorities:

  1. Convert Conventional Industry to dedicated TN installations. Per Section 2.5.1, CI converts at 20 BP each (vs 120 BP to build new). Priority order:
    • Construction Factories (to build more installations faster)
    • Mines (to feed your construction)
    • Fuel Refineries (you’ll need fuel for ships)

  2. Research propulsion — you need engines before you can build useful ships. See Section 7.4 for the engine tech tree.

  3. Don’t neglect maintenance — per Section 14.2.3, Maintenance Facilities produce MSP at 20/year base rate. Build some early or your future fleet will deteriorate.

Player: How many construction factories should I aim for?

AI: There’s no fixed answer, but community convention suggests:

  • 400-800 for a comfortable pace
  • 1000+ for aggressive expansion

Per Section 6.3, each CF produces 10 BP/year. With 500 CFs, you produce 5,000 BP/year — enough to build roughly 40 mines or a small warship annually.

Player: I converted 200 CI to construction factories. Now my mineral stockpiles are dropping fast. Is that normal?

AI: Yes — each conversion costs minerals (Section 2.5.1 table). Construction Factories cost 10 Duranium + 10 Neutronium per conversion. With 200 conversions, you spent 2,000 Duranium and 2,000 Neutronium.

Check your mineral stockpiles in the Economics window. If you’re running low, prioritize converting some CI to Mines next, or pause conversions until stockpiles recover.

This coaching pattern — situation, guidance, follow-up, adjustment — represents how AI can serve as a knowledgeable companion throughout your Aurora journey.

1.4.6.2 Mid-Game Crisis: NPR Invasion

This example shows AI coaching during an urgent situation:

Player: HELP. An NPR fleet just jumped into my home system. 12 ships, I have 4 destroyers. What do I do RIGHT NOW?

AI: Don’t panic. Immediate actions:

  1. Pause and assess — Click the smallest time increment. How far are they from Earth? What are their ship classes?

  2. Check their stance — Are they already hostile or just present? Look at the Diplomacy window. If they haven’t fired, you may have time.

  3. If hostile and close:
    • Set your destroyers to maximum military speed (burns fuel fast but you need distance)
    • Retreat toward Earth if you have PDCs, or away if you don’t
    • DO NOT engage 4v12 unless they’re clearly inferior tech
  4. If you have time (they’re far out):
    • Query database: What are your PDC defenses on Earth?
    • Can you rush-build missiles from ordnance factories?
    • Are there any other ships in nearby systems you can recall?

Player: They’re 200 million km out, heading toward Earth at 2000 km/s. I have 2 PDCs (Planetary Defence Centres) with lasers.

AI: You have roughly 28 hours game-time. That’s workable.

  1. Don’t intercept with your destroyers — they’ll die. Use them as scouts at maximum sensor range.

  2. Position destroyers to track the NPR fleet — you need to know their exact approach vector.

  3. Your PDCs are your main defense. Per Section 12.7, PDCs don’t need fuel or maintenance. They’ll fight until destroyed.

  4. Check your PDC designs — what’s their range? If the NPR has missiles and you only have beam PDCs, you need to consider evacuation planning.

Player: PDCs have 200k km range lasers. NPR ships look like they have missile launchers.

AI: That’s a problem. Missile range typically exceeds 200k km. They may be able to stand off and bombard.

Options:

  1. Hope your PDCs survive long enough — fortification level matters (Section 12.7.2). What’s Earth’s fortification?
  2. Evacuate critical personnel — do you have any freighters? You might not save the colony but you can save commanders.
  3. Diplomatic Hail Mary — try opening communications. Some NPRs will negotiate if you catch them before combat starts.

This is a survival situation. Learn from it for next game: always have mobile missile ships, not just beam PDCs.

Crisis coaching requires faster responses with less verification. The AI provides options and trade-offs rather than optimal solutions, because in Aurora, there often isn’t one right answer — only choices with consequences.

1.4.7 Trusting AI Output

Updated: v2026.02.01

AI assistants can make mistakes. This section covers how to verify AI responses, understand AI limitations, and how this manual is designed to minimize errors.

1.4.7.1 Verification Habits

Develop these habits when using AI coaching:

Demand Citations

Always ask the AI to cite manual sections:

What is the formula for population growth?
Cite the specific manual section and reference number.

If the AI cannot cite a source, treat the answer as unverified.

Cross-Check Critical Values

For important numbers (costs, formulas, tech requirements), verify against:

  • The manual’s database references
  • The in-game tooltips
  • The actual game database if you’re comfortable with SQLite queries

Distinguish VB6 from C#

Aurora C# differs significantly from the older VB6 version. If an AI response mentions mechanics that seem unfamiliar, ask:

Is this mechanic from VB6 Aurora or C# Aurora?
This manual covers C# only.

Test Before Committing

For strategic advice, test on a small scale before full commitment:

  • Build one ship of a new design before queuing ten
  • Try a tactic with a single fleet before your entire navy
  • Save before major decisions so you can revert if needed

1.4.7.2 AI Limitations

Understand what AI cannot do:

  • No direct game access: AI cannot see your game state unless you describe it or use database queries
  • No real-time updates: AI doesn’t know what happened since you last described your situation
  • Potential hallucination: AI may confidently state incorrect game mechanics — always verify critical information
  • No substitute for experience: Reading about missile combat is not the same as fighting a missile duel

AI assistance accelerates learning but does not replace playing the game.

1.4.7.3 How This Manual Prevents Hallucination

A common concern with AI-assisted learning is “hallucination” — AI confidently stating incorrect information. This manual is specifically designed to minimize this problem when used as an AI knowledge base.

Database-Verified References

Every factual claim in this manual includes a numbered reference linking to its source. The primary source is the Aurora game database itself (AuroraDB.db), with citations specifying exact table names, column names, and values:

Maintenance Facilities produce MSP at a base rate of 20 per year [3].

[3]. Aurora C# game database (AuroraDB.db v2.7.1) — FCT_TechSystem
     TechTypeID=209, AdditionalInfo=20.0, TechDescription="Annual
     production rate for a single maintenance facility"

When an AI cites this manual, it is citing database-verified facts, not forum posts, wiki summaries, or community folklore.

Explicit Uncertainty Marking

Claims that cannot be verified against authoritative sources are explicitly marked:

*(unverified — #837, requires live testing)*

This marking warns human readers and signals to AI that this specific claim should be treated with lower confidence.

Structural Separation

The manual separates different types of content:

Content Type Indicator AI Should Treat As
Database-verified facts \hyperlink{ref-X.Y-N}{[N]} citation High confidence
Unverified claims *(unverified — #NNN)* marker Low confidence, recommend verification
Community recommendations “Tip:” or “community convention” language Opinion/guidance, not game mechanics
Editorial content Explicitly labeled sections Author perspective, not game facts

Why This Matters

Traditional game documentation sources are problematic for AI:

  • Forum posts mix accurate information with outdated advice, speculation, and version-specific mechanics
  • Wiki articles may describe VB6 Aurora mechanics that no longer apply to C# Aurora
  • YouTube tutorials become outdated as the game updates
  • AI training data includes all of the above without distinguishing quality

This manual addresses these problems by:

  1. Version pinning: All references cite the specific database version (currently v2.7.1)
  2. C# focus: The manual explicitly covers only C# Aurora, eliminating VB6 confusion
  3. Active maintenance: Unverified claims are tracked via GitHub issues and resolved over time
  4. Testable claims: Database references can be independently verified by anyone with SQLite access

1.4.7.4 Verification Workflow

If you suspect an AI response based on this manual might be incorrect:

  1. Ask for the citation: “Which manual section and reference number supports that claim?”
  2. Check the reference: Look up the cited reference in the manual’s References section
  3. Verify against the database: If you have SQLite access, run the query yourself
  4. Check for unverified markers: If the claim has *(unverified)*, treat it as provisional
  5. Report discrepancies: File a GitHub issue if you find an error — the manual improves through community verification

This workflow transforms AI-assisted play from “trust the AI” to “trust but verify” — with verification made practical by the manual’s citation infrastructure.

1.4.8 Frequently Asked Questions

Updated: v2026.02.01

What if my AI can’t access files?

Not all AI interfaces support file system access. Alternatives:

  1. Copy-paste relevant sections — When asking about a specific topic, paste the relevant manual section into the conversation
  2. Link to the GitHub repository — Some AI tools can fetch web content from github.com/ErikEvenson/aurora-manual
  3. Use the PDF — Upload the manual PDF if your AI supports document uploads
  4. Summarize your question with context — “According to the Aurora manual, maintenance facilities produce MSP. Can you explain how the production rate scales with technology?”

How do I find my database path?

The Aurora database location depends on your installation:

Platform Typical Path
Windows C:\Games\Aurora\AuroraDB.db or wherever you extracted Aurora
macOS (via Wine/Crossover) ~/Library/Application Support/CrossOver/Bottles/[bottle]/drive_c/Aurora/AuroraDB.db
Linux (via Wine) ~/.wine/drive_c/Aurora/AuroraDB.db

If you can’t find it, search your system for AuroraDB.db. There should be only one file with this name.

The AI gave me advice that seems wrong. What do I do?

  1. Ask for the citation — “Which manual section supports that claim?”
  2. If no citation exists — The AI may be drawing from general training data (forums, wikis) rather than this manual. Ask it to re-answer using only the manual.
  3. If the citation exists but seems wrong — Check the manual’s References section for that citation. The database query is listed there.
  4. If you find an actual error — File a GitHub issue. The manual improves through community verification.

Can AI help me cheat or exploit the game?

AI can explain any mechanic, including ones that might be considered exploitative. Whether to use that knowledge is your choice. Aurora is single-player (or cooperative multiplayer) — there’s no competitive harm in optimizing your play. Many “exploits” are simply deep understanding of intended mechanics.

How do I get AI to remember my game state across sessions?

Most AI conversations don’t persist between sessions. Options:

  1. Keep a game journal — Maintain a text file with your current state that you paste at the start of each AI session
  2. Use AI with memory features — Some AI tools offer conversation memory or project context
  3. Rely on database queries — If your AI can query AuroraDB.db, it can rediscover your game state each session
  4. Use /coach — The coach skill queries your database automatically each time

The AI is hallucinating mechanics that don’t exist. Why?

The AI may be:

  1. Confusing VB6 and C# Aurora — Many mechanics changed between versions. Explicitly state “C# Aurora only”
  2. Drawing from forum speculation — Ask it to cite manual sections, not general knowledge
  3. Inventing plausible-sounding mechanics — Aurora is complex enough that fabricated mechanics can sound real. Always verify critical claims.

1.4.9 Multiplayer Considerations

Updated: v2026.02.01

Aurora supports multiplayer via hotseat (players take turns at the same computer). AI coaching in multiplayer contexts requires additional considerations.

Information Boundaries

In competitive multiplayer, each player should only access information about their own faction:

  • Don’t query other players’ data — Even if the AI can access the full database, avoid queries that reveal opponent ship designs, fleet positions, or research progress
  • Faction-specific queries — Filter database queries by your RaceID to see only your own data
  • Honor the fog of war — If your sensors haven’t detected something, don’t ask the AI about it

Example faction-filtered query:

SELECT Name, Cost FROM FCT_ShipClass WHERE RaceID = [YourRaceID];

Note: The /coach skill automatically filters to your race only (NPR=0), making it safe for multiplayer use.

Coaching Etiquette

In multiplayer games, agree on AI coaching rules upfront:

Approach Description
No AI Pure human play; AI assistance considered cheating
Mechanics only AI can explain rules but not provide strategic advice
Full coaching AI assistance permitted for all players equally
Asymmetric New players may use AI; experienced players may not

There’s no “right” answer — multiplayer Aurora is about the experience you and your partners want.

Turn Time Considerations

AI consultation takes time. In multiplayer:

  • Set turn time expectations — Agree on reasonable turn lengths
  • Prepare questions in advance — Think about what you’ll ask during opponents’ turns
  • Don’t abuse pauses — Extended AI consultation during your turn can frustrate other players

Cooperative Multiplayer

In cooperative games (humans vs. NPRs), AI coaching benefits everyone:

  • Share AI insights openly
  • Use AI to coordinate strategy between factions
  • Query the database together to plan joint operations

Cooperative Aurora with AI coaching can feel like having a shared advisor helping your alliance succeed.

1.4.10 Community Contribution

Updated: v2026.02.01

This manual improves through community verification. You can help:

Verifying Unverified Claims

Claims marked *(unverified — #NNN)* need testing. To help:

  1. Browse open issues at github.com/ErikEvenson/aurora-manual/issues
  2. Filter by the unverified label to see claims needing verification
  3. Test in-game — Run the experiment described in the issue
  4. Report results — Comment on the issue with your findings, including:
    • Aurora version tested
    • Steps you took
    • Results observed
    • Any database queries you ran

Reporting Errors

If you find an error in the manual:

  1. Search existing issues to avoid duplicates
  2. Create a new issue with:
    • The section and line containing the error
    • What the manual claims
    • What the game actually does
    • How you verified (database query, in-game test, etc.)

Contributing Content

The manual accepts contributions via GitHub pull requests. See the repository README for contribution guidelines. Priority areas:

  • Screenshots of UI elements referenced in the manual
  • Verification of *(unverified)* claims
  • Corrections to outdated information after Aurora updates

Every verified claim makes AI-assisted play more reliable for all players.

Back to top

Aurora 4X Manual & Guide - Unofficial community documentation for Aurora C# (game by Steve Walmsley)

This site uses Just the Docs, a documentation theme for Jekyll.