A cross-platform command center for the machine that agents drive the same way you do — as an installed agent skill, or over MCP. Provides a unified CLI over windows, input, screenshots, files, shell commands, and packages on macOS, Windows, and Linux.
aos exposes only 2 tools with minimal schema definition. The 'commands' tool has a basic boolean-parameter schema, but the 'version' tool has an empty input schema. Descriptions are present but generic (e.g., 'Print the version of aos' is 26 chars, below the 50-char baseline for LLM optimization). No output schemas are documented. Tool definitions appear inferred from test/dispatch files rather than explicit MCP registration, capping per-tool scores at 50. No error handling guidance, no parameter validation hints, and no composition support for chaining tools.
List all available commands, optionally filtered by platform and display availability. Supports JSON output and validation checking.
Print the version of aos
Tool definitions inferred from test/dispatch files, not explicitly registered in MCP server code. No visible tool registration with MCP framework.
No output schemas documented for either tool. LLMs cannot plan downstream calls or extract structured data.
Descriptions are generic and under 50 chars (baseline for LLM optimization). 'Print the version of aos' lacks context on when/why to call it.
No error handling guidance. Tools do not indicate what to do if a command is not found or version resolution fails.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 42 | 2026-07-28+ | v2 |