hlv CLI
A Rust binary that checks agent-written code against the spec: 30-plus validations of contracts, test specs and gates.
- 81 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- Runs code and writes project files under `.hlv/` and skills directories.
- The MCP server exposes 29 tools and project resources to clients.
Install
- Works in
Manual install
curl -fsSL https://github.com/lee-to/hlv/releases/latest/download/hlv-aarch64-apple-darwin.tar.gz | sudo tar xz -C /usr/local/binThe Apple Silicon macOS variant from the README, other systems have their own commands. The command unpacks a binary with administrator rights.
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Runs code and writes project files under `.hlv/` and skills directories.
- The MCP server exposes 29 tools and project resources to clients.
This is third-party code. Review the repository files before installing.
What it does
hlv is a compiled Rust utility that calls no model and generates no code. A human states requirements and constraints in declarative files, an agent writes code from them, and `hlv check` verifies the chains from requirement to contract, test and gate and finds divergence. `hlv init` scaffolds the project and installs six skills: `/hlv-artifacts`, `/hlv-generate`, `/hlv-implement`, `/hlv-validate`, `/hlv-verify` and `/hlv-questions`. `hlv init --adopt` attaches the tool to an existing project, and `hlv mcp` starts an MCP server for agents and web dashboards. All commands accept `--json`.
Who it is for. Teams that hand code writing to an agent and want automatic checks of the result against contracts and gates.
Good fit when
- An agent implements a plan stage and all mandatory gates must pass before moving on.
- Specs are introduced into an existing Laravel, Go, Node, Python or Rust project with `hlv init --adopt`.
- CI needs a strict check: `hlv check --strict` turns warnings into errors.
- A web dashboard or Claude Code should read project state through the `hlv mcp` MCP server.
Not a fit when
- You need a quick edit with no specs and contracts.
- You need a tool that generates code itself: hlv only validates.
- An unstable format is unacceptable: the author warns of breaking changes between releases.
Example request
Run `hlv check` and show which contracts are not linked to code and which gates have not passed.Limitations
The project is under active development and the author warns about breaking changes between releases. The method requires writing contracts, test specs and gates up front, which is a discipline of its own and overkill for quick edits. Skills are installed by `hlv init`, and on Windows the binary must be put on `PATH` manually. Installing from a release writes the binary to `/usr/local/bin` with administrator rights. For simple cases the author points to a lighter project, AI Factory.
How to disable. Remove the `hlv` block from the MCP configuration and delete the `hlv-*` skills from the agent skills directory. The binary is removed from `/usr/local/bin`, and the `.hlv/` directory in the project can be deleted if specs are not needed.
FAQ
How do I install hlv on Apple Silicon macOS?
Download and unpack the release: `curl -fsSL https://github.com/lee-to/hlv/releases/latest/download/hlv-aarch64-apple-darwin.tar.gz | sudo tar xz -C /usr/local/bin`. The README has commands for other systems, and `cargo install --path .` builds from source.
Does hlv call a language model?
No. The utility only validates, and the agent writes code from the spec.
How do I connect the hlv MCP server?
Add a server to the client configuration with command `hlv` and arguments `mcp`, `--root`, `/path/to/your/hlv-project`. For web clients there is `hlv mcp --transport sse --port 3000`.
Which skills does `hlv init` install?
Six: `/hlv-artifacts`, `/hlv-generate`, `/hlv-implement`, `/hlv-validate`, `/hlv-verify` and `/hlv-questions`. They carry a project from the interview to stage-by-stage validation.
How do I update hlv?
`hlv update` updates the binary and synchronizes managed project files, including agent skills.
Discussion
No questions or reviews yet.
Related
CLI toolsA static linter for agent instructions: finds ambiguous tool descriptions and schema gaps before a run, with no model or network
A plugin that teaches agents to write Hegel property-based tests for six languages and review them for weak spots
Reviews the git diff and the conversation with the agent: catches bugs and gaps between your request and the agent's result. CLI, skill, GitHub Action
A local debugger for TypeScript agents: run step trees, CI trajectory checks and portable evidence bundles, with no account.
A CLI that rates the maturity of an AI agent harness in a repository from L0 to L4: 36 file checks with no model calls or network.
A CLI that finds traces coding agents leave in code, like filler comments, swallowed errors and fake imports, and scores it 0–100
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| hlv (this card) | no | 81 | |
| LeanSpec | no | 302 | also spec-driven, with short specs and a task board; hlv focuses on contract and gate validation by a binary |
| Autospec | no | 148 | drives work from description to implementation with YAML artifacts; hlv separately checks requirement traceability |
README in short
Three layers of responsibility: the human states the what, the agent writes the how, hlv checks the proof. Then come sample `hlv check` output, a table of 30-plus validations, the command list, MCP and per-platform installation.