The agent-native standard
One bar for agent-readiness, measured on two surfaces: your CLI and your website. Scored MUST / SHOULD / MAY, not asserted.
See where things stand
Curated CLIs, ranked by credit-weighted agent-readiness.
Public sites, ranked by global agent-readiness.
Scored against the 8 principles. Run anc audit <tool> locally for source + project depth. Full board ▸
Scored against the emerging agent-web standards: MCP, llms.txt, OpenAPI, JSON Schema, discovery. anc audits; it doesn't own them. Full board ▸
or try , , , , , or .
Eight principles for a CLI
The standard anc.dev authors. Each is a testable contract with a MUST / SHOULD / MAY obligation and a named failure mode.
Five checks for a website
Not a standard anc owns; an audit of your agent-facing surface against what the ecosystem is converging on.
- P1
Every automation path MUST run without human input. A CLI tool that blocks on an interactive prompt is invisible to an agent: the agent hangs, the user sees nothing, and the operation times out silently.
- P2
CLI tools MUST separate data from diagnostics and offer machine-readable output formats. Mixing status messages with data forces agents into fragile regex extraction that breaks on any format change.
- P3
Help text MUST be layered so agents (and humans) can drill from a short summary to concrete usage examples without reading the entire manual. The critical layer is the one that appears after the flags list, because that is where readers look for invocation patterns.
- P4
CLI tools MUST detect invalid state early, exit with a structured error, and tell the caller three things: what failed, why, and what to do next. An error that says "operation failed" gives an agent nothing to act on.
- P5
Every CLI with write operations MUST support --dry-run so agents can preview a mutation before committing it. Commands MUST make the read-vs-write distinction visible from name and --help alone, and destructive writes MUST require explicit confirmation. An agent that cannot distinguish a safe read from a dangerous write will either avoid the tool or execute mutations blindly. Both are failure modes.
- P6
CLI tools MUST integrate cleanly with pipes, scripts, and other tools. That means handling SIGPIPE, detecting TTY for color and formatting decisions, supporting stdin for piped input, and maintaining a consistent, predictable subcommand structure.
- P7
CLI tools MUST provide mechanisms to control output volume. Agent context windows are finite and expensive: a tool that dumps tens of thousands of lines of unfiltered output wastes tokens on every request and can exceed smaller context windows entirely, breaking the conversation that invoked it. "High-signal" here means the bytes that survive --quiet are the ones the caller asked for (data and errors), not progress, decoration, or chatter.
- P8
Without a skill bundle, every fresh agent invocation begins the same way: pull --help, infer the idioms, try a command, parse the error, try again. A skill bundle (canonical names AGENTS.md or SKILL.md) collapses that loop: agent-discoverable through filesystem convention rather than through --help, loaded once, recognized thereafter.
- C1DiscoverabilityMUST
robots.txt,sitemap.xml, Link headers, DNS-AID under_agents. - C2
llms.txt,Accept: text/markdown, JSON-LD, semantic landmarks. - C3Bot & crawl policySHOULD
AI-crawler rules, Content-Signal,
security.txt, Web Bot Auth. - C4MCP & APIMUST
Initialize handshake,
tools/list, error codes,.well-knowncard, OpenAPI. - C5
A2A agent card, agent-skills index, OAuth discovery,
auth.md.
To audit a CLI, install the linter (anc) locally: see /install. To audit a website and its MCP server, use /web-audit or the audit_website MCP tool. To install the agent-native-cli skill bundle (the Claude Code / Codex / Cursor / OpenCode skill), see /skill. Agents that speak MCP can connect directly to the streamable-HTTP server at https://anc.dev/mcp. See /mcp-skill for the wire contract and tool catalog.