Skip to main content
Stoffel Developer Skills are AI-agent-ready playbooks for developers building applications with the Stoffel framework. They are designed to help coding agents and humans choose the right Stoffel workflow, use current CLI and SDK APIs, run local MPC smoke tests, generate typed client IO bindings, prepare deployment-shaped network handoffs, and debug common app-level failures. These skills are app-development guides, not internal maintainer guides. They focus on creating and validating Stoffel applications with public tooling.

​
Multi-user application invariant

For client-owned private input, each input owner’s device or process is a distinct participant MPC client. It submits directly through the Stoffel client protocol to the separately deployed MPC network. The application control plane handles public metadata, authorization, session/bootstrap configuration, non-sensitive receipts, lifecycle, and explicitly authorized opened aggregates; it must not receive or persist participant plaintext. A backend gateway that sees raw input is a distinct, weaker trust model and requires explicit approval. For rooms, voting, auctions, private matching, prediction markets, federated analytics, or other multi-user apps, begin with Stoffel Full App Golden Path, then apply Stoffel App Network and Off-Chain Integration. Use the main docs for canonical CLI, SDK, language, and deployment behavior. Use these skills as the agent execution layer: they tell an AI coding agent which docs to consult, what order to work in, what commands to run, and what evidence to collect before reporting that a task is done.

​
Use with AI agents

Mintlify serves these skills through standard skill discovery endpoints and a hosted search MCP server. Use both when possible:
  • Skills tell an agent how to build with Stoffel.
  • MCP gives the agent live read/search access to the current docs.

​
Install the Stoffel skills

Install the published Stoffel skills with the skills CLI:
Useful variants:

​
Connect the Stoffel docs MCP server

Mintlify hosts the Stoffel search MCP server at:
Use the MCP server when an agent needs to search the docs, read full pages as Markdown, or discover the hosted skill.md files as MCP resources. Install it into supported local agents with add-mcp:

​
Agent harness setup

Use the generic setup prompt from Installation first. If your agent needs explicit MCP configuration, add the same Stoffel docs MCP URL through that harness’s MCP setup path. After setup, ask the agent to prove it can search the Stoffel docs before asking it to edit code. After connection, the MCP server exposes read-only tools for searching Stoffel docs and querying the docs filesystem. Tool names are client-specific, but they correspond to:
  • search_stoffel_documentation
  • query_docs_filesystem_stoffel_documentation

​
Discovery endpoints

Agents can discover skills and MCP metadata programmatically:
  • https://docs.stoffelmpc.com/skill.md
  • https://docs.stoffelmpc.com/.well-known/agent-skills/index.json
  • https://docs.stoffelmpc.com/.well-known/skills/index.json
  • https://docs.stoffelmpc.com/.well-known/mcp
  • https://docs.stoffelmpc.com/.well-known/mcp.json
  • https://docs.stoffelmpc.com/.well-known/mcp/server-card.json
  • https://docs.stoffelmpc.com/.well-known/mcp/server-cards.json

​
Verify agent access

After MCP is connected, ask the agent to search the Stoffel docs for ClientStore or local MPC and summarize the relevant command. That verifies the agent can use live docs, not just the installed skill text.

​
Choose the first skill

For runnable app tasks, the agent should return command output from each changed layer. A code diff without stoffel check, stoffel build, and a local MPC run is not enough.

​
Skills

  • Stoffel App Getting Started — Install the Stoffel tooling, create a new app, run first local smoke tests, and choose the right development path.
  • Stoffel CLI App Workflow — Use the stoffel CLI to init, check, build, compile, run, test, inspect, and troubleshoot Stoffel apps.
  • Stoffel Full App Golden Path — Build a complete app from privacy boundary through StoffelLang, Rust SDK integration, local MPC validation, typed bindings, and deployment handoff.
  • Stoffel-Lang App Programming — Write .stfl application logic using supported Stoffel-Lang syntax, types, builtins, and example patterns.
  • Stoffel Secret MPC Programming — Build MPC apps with secret types, Share, ClientStore, Mpc, MpcOutput, and runnable private-input examples.
  • Stoffel Rust App SDK — Embed Stoffel in Rust apps using the SDK for compilation, bytecode loading, local execution, clients, and servers.
  • Stoffel Typed Client IO Bindings — Generate and use Rust typed client input/output bindings from exact Stoffel bytecode manifests.
  • Stoffel Local MPC Dev Loop — Run local MPC smoke tests, ClientStore input flows, hot reload, and SDK local coordinator-backed execution.
  • Stoffel App Network and Off-Chain Integration — Move from local bytecode to client/server builders, network config, and off-chain coordinator integration.
  • Stoffel Deployment Runbook — Prepare production-shaped artifacts, topology, coordinator settings, identity material, client config, and operational verification.
  • Stoffel App Troubleshooting — Diagnose app-level init, check, build, run, local MPC, binding, SDK, and network failures with evidence.
  • Stoffel AI Agent Implementation — Give AI coding agents backend, input/output, validation, and cost-model context before they write Stoffel code.

​
Source and maintenance

The visible pages are mirrored into .mintlify/skills/<slug>/SKILL.md so developers can read them in the docs and AI agents can fetch them as skill resources. After editing a page in developer-skills/, run python scripts/sync_developer_skills.py before validating.