STS2 AI Agent
Enables AI agents to play Slay the Spire 2 by exposing game state and actions through an MCP server, supporting combat, rewards, and run management.
README
STS2 AI Agent
https://github.com/user-attachments/assets/89353468-a299-4315-9516-e520bcbfbd4b
中文版说明请见 README.zh-CN.md.
STS2 AI Agent is a Slay the Spire 2 mod + MCP server bundle:
STS2AIAgent: exposes game state and actions through a local HTTP APImcp_server: wraps that local API as an MCP server for AI clients
Detailed MCP tool documentation lives in mcp_server/README.md. If you want an agent workflow on top of it, start with skills/sts2-mcp-player/SKILL.md.
Quick Start
1. Install The Mod
After downloading and extracting the release package, copy these files into your game's mods/ directory:
STS2AIAgent.dll
STS2AIAgent.pck
mod_id.json
The default Steam install path is usually:
C:\Program Files (x86)\Steam\steamapps\common\Slay the Spire 2
Your final layout should look like this:
Slay the Spire 2/
mods/
STS2AIAgent.dll
STS2AIAgent.pck
mod_id.json
2. Start The Game And Confirm The Mod Is Loaded
Launch the game normally so the mod can load with it.
Then open:
http://127.0.0.1:8080/health
If the endpoint responds, the mod is running.
3. Start The MCP Server
Prepare the environment first:
- Install
Python 3.11+ - Install
uv
Install uv on Windows:
powershell -ExecutionPolicy Bypass -c "irm https://astral.sh/uv/install.ps1 | iex"
On macOS:
brew install uv
Then start the default stdio MCP server.
Windows:
powershell -ExecutionPolicy Bypass -File ".\scripts\start-mcp-stdio.ps1"
macOS / Linux:
./scripts/start-mcp-stdio.sh
This is the recommended default. Most desktop AI clients prefer stdio MCP integration.
4. Connect Your MCP Client
If your client supports command-based MCP startup, point its working directory at mcp_server/ and use:
uv run sts2-mcp-server
If your client works better over HTTP, start the network server instead.
Windows:
powershell -ExecutionPolicy Bypass -File ".\scripts\start-mcp-network.ps1"
macOS / Linux:
./scripts/start-mcp-network.sh
Default MCP endpoint:
http://127.0.0.1:8765/mcp
What The Project Can Do
The current main branch provides a playable MCP integration for STS2, including:
- reading live game state
- listing currently legal actions
- driving combat, rewards, shops, map routing, events, rest sites, chests, capstone selection, and bundle selection
- enriched combat and run payloads (Ascension, act/boss ID, enemy/move ID) for AlphaZero training
resolve_rewardsatomic action for controlled reward resolution- reducing polling through SSE events
- exposing MCP over
stdioor HTTP - serving live game metadata for cards, relics, monsters, potions, and events via the Mod API
- supporting layered planner / combat agent handoff flows
increase_ascension/decrease_ascensioncontrols in character select
See mcp_server/README.md for the detailed tool surface.
FAQ
http://127.0.0.1:8080/health Does Not Open
Check these first:
- The game is actually running
STS2AIAgent.dll,STS2AIAgent.pck, andmod_id.jsonare all inside the game'smods/directory- The files were not duplicated or renamed by the OS
- You copied them into the Steam game directory, not the repository directory
The MCP Server Starts But Cannot Read Game State
That usually means mcp_server is running, but the in-game mod is not connected. Confirm:
- The game is running
http://127.0.0.1:8080/healthis reachable- The MCP server is still pointing at
http://127.0.0.1:8080
Should I Enable Debug Actions?
Usually no.
Developer-only actions such as run_console_command are disabled by default and should stay disabled in normal use and releases.
Building From Source
If you are building from source instead of using a release package:
Windows:
powershell -ExecutionPolicy Bypass -File ".\scripts\build-mod.ps1" -Configuration Release
macOS / Linux:
./scripts/build-mod.sh --configuration Release
More complete environment, path-discovery, and validation notes are in build-and-env.md.
Repository Layout
STS2AIAgent/: game mod sourcemcp_server/: MCP server sourcescripts/: startup, build, and validation scriptsdocs/: supporting documentationskills/: companion skills
License
This project is licensed under the GNU Affero General Public License v3.0 only (AGPL-3.0-only).
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.