brettdavies

Agentnative — AI skill for Claude Code

AI community

The agent-native CLI standard — eight principles for CLI tools operated by AI agents.

How to install Agentnative

This entry records only its repository, not the path inside it, so there is no exact command to give. Open brettdavies/agentnative and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Agentnative does

The agent-native CLI standard — eight principles for CLI tools operated by AI agents.

Alternatives in AI

Opengap — A framework-agnostic, git-native standard for defining AI agents 2.9k ★
Codeseek — Rust-powered code intelligence CLI for AI coding agents 764 ★
Pebrel — AI-native, GPU-accelerated terminal emulator for Windows with SSH, persistent sessions, split panes, and first 670 ★

README

The Agent-Native CLI Standard

Eight principles for CLI tools operated by AI agents.

AI agents operate CLIs differently than humans do. They can't answer interactive prompts, can't parse vague output, and can't recover from errors that don't say what to do next. The principles below define what *agent-native* means in RFC 2119 language, with machine-readable `requirements[]` so an auditor (and graders) can pin against the standard.

**Early-stage spec.** Ideas, debate, and contributions are welcome: pressure-tests against any principle, real-CLI grading submissions, counter-examples that argue a tier sits wrong, or new principles that the genre is missing. The spec is published as `status: active` because the contracts are stable enough to cite, not because anything here is locked. See [`CONTRIBUTING.md`](CONTRIBUTING.md) for how to file each type.

The four artifacts

Spec: this repo. Eight RFC 2119 principles plus machine-readable requirements[] in YAML frontmatter. Currently v0.5.0; all principles ship as status: active.
Linter: `anc`. Scores any CLI repo against the spec. Pins against requirement IDs, not prose.
Skill bundle: `agentnative-skill`. Agent-facing guide that teaches agents how to invoke anc and remediate findings against the spec; vendors the principles for offline use.
Leaderboard: anc.dev/scorecards. Top CLIs graded live; submit a PR to add yours.

Quick start

brew install brettdavies/tap/agentnative
anc audit .

Also installable via `cargo install agentnative` or platform-specific archives on [GitHub Releases](https://github.com/brettdavies/agentnative-cli/releases).

Run `anc audit . --output json` for machine-readable scorecards. Per-principle filtering via `anc audit . --principle <1-8>`. `anc emit schema` prints the scorecard JSON Schema (draft 2020-12) for