The intersection of CLI tooling and AI agents just got a practical upgrade with the release of Oriko, a command-line interface designed to generate tailored CVs for job applications. Developed by Serkan Sipahi and hosted on GitHub, Oriko allows AI assistants to read a user's base CV, analyze a specific job posting, and produce a customized version that highlights relevant skills while omitting extraneous details. The tool is built to be driven entirely by agents, offering a schema and skill set that integrates with Claude Code and other frameworks, effectively turning the terminal into a resume customization engine.
Agentic Meets Deterministic Design
Oriko operates on a hybrid architecture that separates judgment from execution. The AI agent handles the creative and analytical workβreading job descriptions, selecting relevant experience, and reordering skillsβwhile the CLI enforces strict structural rules. The underlying skill is defined as a state machine with fixed steps: the agent must read the posting, inspect the CV, draft a version, and check it. If a draft fails validation, such as a block overflowing a page, the agent is forced to correct it within three attempts before asking the human for intervention. This prevents the common LLM pitfall of hallucinating facts or generating broken layouts.
Terminal-First Preview and Agent DX
A standout feature is the terminal preview, which renders the CV layout in ASCII art directly in the shell, showing page breaks and fill percentages before any file is generated. This aligns with the project's adherence to "Agent DX" rules, inspired by Justin Poehneltβs 2026 article on rewriting CLIs for AI agents. Oriko accepts raw JSON input, provides schemas at runtime, and returns JSON when piped, ensuring that agents can parse errors and next steps without guessing. The CLI refuses unknown commands and options silently, forcing the agent to act only on explicit, valid inputs.
Installation and Compatibility
Installation is handled via a curl script that sets up both the CLI and the agent skill, with optional flags for PDF generation via LibreOffice. The tool stores data in ~/.oriko, keeping CVs and job-specific profiles separate from the code repository. While primarily tested with Claude Code, the skill is designed to work with any agent capable of executing commands. The project is licensed under the GNU Affero General Public License v3.0, ensuring that any modified versions distributed over a network must also publish their source code.
Key Takeaways
- Oriko combines AI judgment with deterministic CLI checks to prevent CV layout errors and factual hallucinations.
- The tool features a real-time terminal preview of the CV, allowing agents and users to verify formatting before export.
- Installation integrates the agent skill automatically for Claude Code and other supported environments.
- The project follows specific "Agent DX" rules, including raw JSON input and schema introspection, to streamline agent interaction.
The Bottom Line
Oriko is a smart move for the agent ecosystem, proving that mundane tasks like resume formatting benefit immensely from deterministic guardrails. Itβs a welcome departure from fragile, prompt-only workflows.