Experimental reference implementation for the MaxOS specification. A board-driven virtual employee system with MCP server integration for skill management, API calls, memory operations, and workflow automation.
MaxOS reference implementation has 7 tools with inconsistent quality. Tool names follow verb_noun convention (list_skills, get_skill_page, api_call, memory_read_scoreboard), which is good. However, descriptions are minimal (10-50 chars), parameter descriptions are sparse or missing, and input schemas lack proper type definitions and constraints. The api_call tool accepts a generic 'object' for body without field documentation. No output schemas are documented. Error handling is absent, tools provide no recovery guidance. This is a reference implementation, not production-grade.
Execute a scoped API call to an external service with credential brokering
Get API reference documentation for a specific service
Get detailed information about a specific skill
Generate a joke based on an initial idea
List available skills for the current employee
Read the current scoreboard from memory
api_call tool accepts 'body' as generic object with no field schema or documentation. LLMs cannot infer what fields are required or valid.
All tool descriptions are under 50 characters and lack WHEN/WHY context. 'List available skills for the current employee' does not explain when to call this vs other discovery tools or what structure is returned.
No output schemas documented for any tool. LLMs cannot plan downstream calls or extract required fields (e.g., does list_skills return skill_id, skill_name, or both?).
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 56 | 2025-06-18+ | v2 |
Record a competition result in memory
api_call tool has 5 parameters but only 2 have descriptions. 'method', 'path', and 'response_format' lack guidance on valid values, format, or constraints.
No error handling or recovery guidance. Tools provide no indication of what to do if a call fails, whether to retry, or how to self-correct invalid input.