Gennadiyev/STS2MCP — Full agentic runs for Slay the Spire 2.

Description and use

**Brief Summary (2‑3 sentences)**
Gennadiyev/STS2MCP is an open‑source mod for *Slay the Spire 2* that exposes the game’s internal state via a Model Context Protocol (MCP) server, enabling external AI agents to read and act on live gameplay data. The project provides a clean API/SDK and a command‑line interface written in C#, making it straightforward to plug in custom AI or reinforcement‑learning agents.

**Value Proposition**

• **Standardised AI‑to‑tool bridge**: By implementing the MCP, the mod turns a video game into a well‑defined data source and control surface, letting developers test, benchmark, or demo AI assistants on a complex, turn‑based environment without building custom instrumentation.
• **Reusable backend component**: The MCP server can be deployed as a standalone service, so the same protocol can be reused for other games or simulations, accelerating integration work across projects.
• **Rapid prototyping**: With the exposed state (cards, enemies, health, actions) and a simple request/response API, researchers can iterate on model architectures, reward functions, or prompting strategies much faster than hand‑crafting data pipelines.

**Practical Adoption Path**

| Step | Action | Why it matters |
|------|--------|----------------|
| 1️⃣ Clone & build | git clone https://github.com/Gennadiyev/STS2MCP && dotnet build | Verifies the C# toolchain and confirms the mod compiles on your platform. |
| 2️⃣ Install the mod | Drop the compiled DLL into the *Slay the Spire 2* Mods folder and launch the game. | Makes the MCP server available inside the running game. |
| 3️⃣ Run the MCP server | Execute the provided CLI (sts2mcp --port 5000) or start the Docker image if available. | Starts a HTTP/WS endpoint that streams game state and accepts action commands. |
| 4️⃣ Connect your AI | Use the generated OpenAPI spec or the simple SDK (Python, C#, etc.) to subscribe to state updates and send actions. | Allows any language‑agnostic model (LLM, RL agent, rule‑based bot) to interact in real time. |
| 5️⃣ Iterate & test | Record episodes, tweak prompts/rewards, and replay logs via the built‑in replay feature. | Provides a fast feedback loop for model improvement. |
| 6️⃣ Harden for production | Add authentication (TLS, API keys), sandbox the mod in a container, and monitor resource usage. | Mitigates security and stability risks before moving beyond prototypes. |

**Production Readiness Assessment**

• **Maturity**: Medium. The repo is actively maintained (last commit 2026‑05‑12) and has a respectable community signal (335 ★, 56 forks). The codebase is small and focused, which eases auditing.
• **Stability**: Good for internal tools or research prototypes. The MCP server runs as a separate process, limiting impact on the game client, but the mod itself depends on the *Slay the Spire 2* version and may break on future game updates.
• **Operational considerations**:
• **Dependencies** – .NET 6+ runtime; optional Docker image for isolation.
• **Security** – No built‑in auth; you’ll need to front the server with a gateway or VPN for production use.
• **Maintenance** – One primary maintainer; check issue backlog for response times.
• **Readiness checklist before production**:

1. Verify license compatibility with your organization.
2. Add TLS/API‑key protection or run behind a trusted network.
3. Pin the game version and monitor upstream updates.
4. Implement health‑checks and logging for the MCP service.

**Bottom line**
Gennadiyev/STS2MCP offers a ready‑made, standards‑based interface to turn *Slay the Spire 2* into a live AI playground. It is well suited for proof‑of‑concepts, research demos, or internal tooling, and can be hardened for production with modest effort around security, version pinning, and monitoring.

Problem

Gennadiyev/STS2MCP helps connect AI assistants to real tools and data through a standard protocol.

Outcome

Medium: useful for prototypes or internal workflows, with dependency and maintenance checks before production.

Workflow

• connect AI agents to tools
• ship Model Context Protocol servers
• standardize integrations

Evidence

Full agentic runs for Slay the Spire 2. A mod that exposes in-game state, and the MCP server for the mod.

Source