Sley Tools MCP server
Commit Oct 27, 2026Build, run, edit, and review native Sley programs using a command-line toolkit, SDKs, and specialized application adapters.
Also known as sley-lang/tools
- Stars
- 0
- Tools
- —
- Downloads, 30 days
- —
- Latest version
- —
- Last commit
- Oct 27, 2026
- License
- Apache-2.0
What is Sley Tools MCP?
Sley Tools is an MCP server. Build, run, edit, and review native Sley programs using a command-line toolkit, SDKs, and specialized application adapters.
Summary generated from the project's README
About Sley Tools MCP server
Developing and managing native Sley programs is possible through this command-line toolkit and its associated SDKs. Developers use these tools to build, run, edit, and review small programs that execute within the Sley core. The toolkit manages files, inputs, and the connection to the Sley core, allowing for the creation of specialized application adapters and native decision programs.
Users interact with the system via a local setup on Linux x86_64, requiring a separately installed Sley 2.1.1 core. The toolkit provides specific commands like init to create workspaces, run to execute programs, and test to verify behavior against examples. It also includes an Agent Connector to provide local coding agents with scoped MCP tools for inspecting and editing Sley projects.
Beyond the CLI, the ecosystem offers a Bridge for calling programs from Python or TypeScript and a Schema Importer to convert JSON Schema subsets into Sley validators. The system handles SSMC1 semantic graphs as the primary program format, with Rust managing the backend, workspaces, and local RPC.
What you can do with it
- 1Automate release readiness decisionsAn AI client can use the Release Readiness program to evaluate facts about a proposed release and receive an allow decision with supporting reasons.
- 2Integrate Sley into existing codebasesDevelopers can call native Sley programs directly from their existing Python or TypeScript applications.
- 3Empower coding agents with project accessYou can provide a local coding agent with specific tools to inspect, run, and edit a Sley project.
- 4Convert data schemas to validatorsA developer can transform a supported subset of JSON Schema into an editable Sley validator using the Schema Importer.
Written from the project's README. Every use case is checked against a quote from it.
README
View on GitHubSley Tools
Build, run, change, and share small programs that execute in Sley. This repository contains the command-line toolkit, application adapters, ten ready-to-run native programs, and examples that show how to use them.
Example: Release Readiness takes facts about a proposed release and returns an allow decision with reasons. You can run it, test its behavior, change its threshold, and export the resulting native program. The decision runs in Sley; the toolkit handles files, input, and the connection to the Sley core.
Try a native program
RC7 is for Linux x86_64 and runs on the separately installed Sley 2.1.1 core (the current Sley core is 2.1.6; doctor accepts only the 2.1.1 core RC7 was built for). Download and verify the RC7 packages, then run:
CORE="$PWD/sley-2.1.1-linux-x86_64/bin/sley"
TOOLS="$PWD/sley-tools-local/bin/sley-tools"
"$TOOLS" doctor --core "$CORE"
"$TOOLS" init my-first-program --template release-readiness --core "$CORE"
"$TOOLS" run --project my-first-program --input my-first-program/inputs/ready.json
"$TOOLS" test --project my-first-program
init creates and commits a Sley program in a new local workspace. run executes that accepted program. test checks the supplied examples against native execution. See Starter for the edit and export steps.
Choose a tool
| Tool | What it helps you do |
|---|---|
| Starter | Create, run, test, edit, and export a first Sley program from the CLI. |
| Bridge | Call a native Sley program from Python or TypeScript. |
| Agent Connector | Give a local coding agent scoped MCP tools to inspect, run, and edit a Sley project. |
| Rules | Use native decision programs for release and resource checks. |
| Data Pipe | Compare and summarize bounded benchmark data from JSON Lines. |
| Review | See what changed between two native program packages and replay the checks. |
| Playground | Explore a few native examples in a local browser interface. |
| Repair Lab | Practice creating and fixing native programs in eight guided exercises. |
| Schema Importer | Turn a supported subset of JSON Schema into an editable Sley validator. |
| Python Function Importer | Turn a restricted pure Python function into an editable Sley program. |
The native program catalog has ten separate examples, each with an artifact, tests, and a README under programs/. The tool pages above explain the applications and entry points; the program pages explain each native computation.
What runs where
Sley 2 programs are accepted SSMC1 semantic graphs. The JSON graph recipes in this repository are inputs to the builder, not a .sley source language. The selected Sley core validates, commits, and executes the graphs. Rust manages authoring, workspaces, artifacts, and local RPC. Python and TypeScript handle application access and external input/output. Data Pipe comparison, selection, and reduction, for example, each run in native Sley graphs.
You can inspect the graph authoring contract, local RPC contract, and artifact format. The implementation spec records the complete scope and release gates.
Current status
The Linux x86_64 RC7 prerelease targets the public Sley 2.1.1 core, and its release assets passed an offline fresh install and the documented native, SDK, MCP, pipeline, lab and importer examples. sley-tools test checks each program with external assertions against native execution. The 2.1.1 core validates native TestCase candidates but does not yet admit or run them, so test reports native core tests as pending (details).
The Playground is a local optional interface; it is not required for CLI, SDK, or native program use.
Repository layout and license
crates/ holds the shared Rust backend and CLI, packages/ the optional Python, TypeScript, and MCP adapters, programs/ the native examples, apps/ the local Playground and Repair Lab, and actions/ the Review GitHub Action. These tools share one backend and release candidate. Each tool has its own README above; each native program has its own package and README.
Licensed under Apache 2.0. See NOTICE for attribution.
More Sley
- sley-lang/sley: the Sley language, core and
sley-agentworkbench. - sley-lang/examples: thirty small native programs with their authoring records.
- Sley Arena: write a bot in Sley and climb a public ladder.
Sley Tools MCP questions
What operating system is supported for the RC7 prerelease?
The RC7 prerelease is designed for Linux x86_64.