
CLI Printing Press
github.com/mvanhorn/cli-printing-press- Category
- Developer Tools
- Rank
- No. 76Tools index
- Listed in
- #3 Turn an API into a CLI
- Pricing
- Open Source
- Platform
- cli
- Type
- TOOL
- Builder
- @mvanhorn
- GitHub
- 4.7k stars
- Latest release
- v4.32.1
- Date
About
A code generator that creates token-efficient CLIs, MCP servers, and Claude skills for any API or website. It analyzes official docs, community tools, and web traffic patterns to generate Go CLIs with local SQLite storage, compound commands, and agent-optimized interfaces.
What it does
CLI Printing Press turns an API target or browsable website into an agent-oriented command-line package. Its workflow researches competing tools, scaffolds Go code, enriches the source description, regenerates, verifies behavior, checks agent readiness, compares alternatives, and packages the result. Each run can produce a command-line binary, an MCP server, research notes, and verification proofs.
Why it's ranked here
The strongest case is its unusually complete generation workflow. It does not stop after mapping endpoints to commands. It adds review gates, runtime checks, deterministic output fixtures, local data patterns, and publishing support. The verdict is positive for teams willing to adopt its opinionated agent workflow, but the documented preference for Claude Code makes other hosts a less proven choice.
What's good
Generated tools support human-readable terminal tables and machine-readable output when piped. Compact output reduces fields for agent calls, while typed exit codes distinguish authentication, API, and rate-limit failures. Store-backed tools can sync structured data into SQLite, search locally, and answer cross-resource questions that a stateless wrapper cannot. CLI and MCP surfaces share the same client, storage, and authentication model.
Tradeoffs
Installation requires a recent Go toolchain, Node and npm, plus both the generator binary and its skills for the intended workflow. Claude Code is the default and best-tested host; Codex support exists, but nested delegation is disabled inside a Codex sandbox. Browser discovery and live verification introduce more moving parts than spec-only generation. Golden checks protect captured behavior, but the documentation explicitly says they are neither compile checks nor complete coverage.
How to use it well
Use it when an API needs a durable agent-facing interface, especially when local search, compound queries, or both CLI and MCP access matter. Start with the managed pipeline when you need resumable phases and inspectable artifacts; use the faster skill path for an end-to-end session. Keep separate product testing for domain correctness, live credentials, and uncaptured edge cases, because fixture verification only protects recorded contracts.
Technical notes+
README.md defines the installer, skill-first workflow, generated CLI and MCP binaries, SQLite-backed commands, output behavior, and host guidance. go.mod identifies module github.com/mvanhorn/cli-printing-press/v4, Go 1.26.5, Cobra, chromedp, OpenAPI parsing, Bluetooth, YAML, and testing dependencies. docs/PIPELINE.md specifies the nine managed phases and their gates, including generation, review, agent-readiness, comparison, and shipping. docs/GOLDEN.md describes byte-level offline fixtures and explicitly limits what those fixtures prove. docs/CODEX.md explains Codex installation and why nested Codex delegation is disabled inside a Codex sandbox. docs/BLE-PROBE.md documents replay-only defaults plus live BLE builds and smoke tests for macOS and Windows.
Observed
- Primary language
- Go, with module path github.com/mvanhorn/cli-printing-press/v4.
- Toolchain requirement
- Go 1.26.5 or newer; the installer also requires Node and npm for skill installation.
- Install surface
- Shell installer or Go install for the generator binary, plus skills installed through the skills CLI.
- Interfaces
- Generator CLI, generated Cobra CLIs, generated MCP servers, and agent skills.
- Agent hosts
- Claude Code is the default and best-tested skill host; Codex is also documented as an install target.
- Generated artifacts
- Runs can produce two binaries, research documents, verification proofs, and pipeline artifacts.
- Verification structure
- The repository documents unit-oriented checks, byte-level golden fixtures, runtime verification, dogfood checks, and scorecards.
- Device support
- BLE discovery supports replay-only operation by default and optional live builds, with macOS and Windows hardware workflows documented.
Read from README.md, go.mod, docs/DOCS.md, docs/CODEX.md, docs/CURSOR.md, docs/GOLDEN.md, docs/SKILLS.md, docs/RELEASE.md, docs/GLOSSARY.md, docs/PATTERNS.md, docs/PIPELINE.md, docs/ARTIFACTS.md, docs/BLE-PROBE.md, docs/PLUGIN-DEV.md, docs/ATTRIBUTION.md.
What it can do
Generate Go CLI applications from API documentation
API documentation → Go CLI binary with commands mapped to API endpoints
Create MCP servers for external APIs
API specifications and documentation → MCP server implementation
Generate Claude skills from website analysis
Website URLs and web traffic patterns → Claude-compatible skill definitions
Analyze community tools and CLI patterns
Community CLI repositories and tools → Optimized command structure and interface design
Create CLIs with local SQLite storage
Data schema requirements from API analysis → Go CLI with embedded SQLite database functionality
Generate compound command structures
Complex API operations and workflows → Multi-step CLI commands with chained operations
Optimize CLI interfaces for AI agent usage
Standard CLI specifications → Token-efficient, agent-friendly command interfaces
Tags
Tech Stack
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.