Incur is a TypeScript framework for authoring command-line interfaces (CLIs) that are equally usable by human operators and AI agents. It enforces typed argument and option schemas with Zod, produces structured JSON/TOON envelopes, supports fetch-mode and MCP exposure, and can auto-generate SKILL.md files and agent-friendly manifests. Typical use cases include building typed tools, exposing CLIs as MCP servers, and creating agent-discoverable command manifests.
Use Incur when you need a robust, typed CLI that: 1) must be call-able by LLM-based agents (MCP or stdio), 2) requires strict input/output schemas, or 3) should auto-generate skill docs/LLM manifests (--llms). It's ideal for developer tooling, automation scripts, and services where both human and agent consumption are required.
.command(), Cli.create(), fetch-mode, MCP flags, and output formatting guidance are included in the SKILL.md body. The doc shows install commands, examples, and best practices for streaming, middleware, and type generation.Common agent integrations: Claude Code, Cursor, and other agents supporting MCP/stdio tools. The framework explicitly mentions MCP support and manifests usable by agent runtimes.
Incur is a TypeScript CLI framework for building agent-friendly CLIs with Zod schemas, structured output, MCP server generation, and skill file auto-generation. The SKILL.md is a comprehensive, well-written reference document covering commands, options, middleware, streaming, and HTTP serving. No bundled scripts were present to test. The skill is purely documentation — a detailed README for the npm package.
Pure documentation skill — no security concerns. Well-structured SKILL.md with proper frontmatter. Comprehensive coverage of the framework's API. Missing scripts/ and references/ directories that would elevate architecture score.