MINIMAX TEAM RULES

Shared MiniMax / MMX CLI operating rules for XiaoMei, DaLong, YuanQi, and the main coordinator. Trigger this skill whenever the task involves MiniMax, minima...

Security Scan

The skill is an instruction-only team rules document whose requirements and instructions are consistent with its stated purpose of standardizing MiniMax / multi-agent behavior.

Purpose & Capability

Name/description (team rules for MiniMax/multi-agent workflows) align with the instructions: the SKILL.md focuses on rollout, shared source updates, and routing rules. It does not request unrelated credentials or binaries.

Instruction Scope

The instructions explicitly direct the agent to identify affected layers and update a shared source-of-truth (references/agent-rules.md), local guidance, memories, and learnings. This is coherent for a rules/rollout skill but means the agent will read and modify workspace/shared files — the user should be comfortable with that level of file modification.

Install Mechanism

No install spec and no code files are present. Instruction-only skills are low-risk from install mechanisms.

Credentials

The skill requests no environment variables, credentials, or config paths. It mentions use of the mmx CLI when appropriate but does not declare the mmx binary as required; this is an advisory preference, not a hidden credential request.

Persistence & Privilege

always is false and the skill does not request elevated or persistent system privileges. Autonomous invocation is allowed (platform default) but does not on its own raise concerns here.

Assessment

This is a rules-and-rollout document and appears internally consistent. Before installing, confirm: (1) you are happy for the agent to read and edit repository/workspace files (the skill tells the agent to update a shared reference and local guidance); (2) the referenced file references/agent-rules.md exists where you expect it, or decide whether the agent should create/update it; (3) if the agent will call mmx-cli for diagnostics, ensure mmx is installed and any required MMX credentials are configured securely (the skill doesn't request credentials itself). If you want stricter limits, restrict the agent's write access or ask the skill author to make edits manual rather than automatic.

tags:

• minimax
• team
• multi-agent compatibility: openclaw license: MIT

Use this skill whenever the work touches any of these topics:

• MiniMax capability selection or mapping
• mmx / mmx-cli install, auth, status, quota, or diagnostics
• Token Plan model availability or remaining quota
• OpenClaw tool ↔ MiniMax native CLI mapping
• quota interpretation / model usage explanation
• multi-agent routing, producer-direct delivery, or “谁来发给阿辉”
• configuration / standards / multi-agent workflow changes that should land consistently across multiple layers

Core Rule

For configuration / standards / multi-agent workflow changes, do one-pass rollout by default:

1. verify the fact,
2. update the shared source of truth,
3. update affected local guidance or references,
4. record the learning if a misunderstanding or correction occurred.

Do not stop at a partial fix when the issue clearly affects multiple layers. Do not make the user push for each missing layer one by one.

Required Execution Pattern

When this skill triggers, default to this checklist:

1. Identify all affected layers (shared docs, local agent guidance, memory, learnings).
2. Update the shared source of truth first.
3. Convert repeated local rule blocks into references to the shared source, keeping only role-specific additions locally.
4. Keep all four roles aligned: 小美 / 大龙 / 元气 / 小语.
5. When explaining quota, use the documented house interpretation from the reference file. Do not guess from field names alone.
6. When finalizing, give the user the outcome, not internal rollout noise.

Shared Source of Truth

Read this reference when you need the concrete rules:

• references/agent-rules.md

Operating Defaults

• Prefer OpenClaw first-class tools for normal user work.
• Use mmx when you need MiniMax-native verification, auth/quota diagnostics, or a capability not directly exposed the way you need.
• Producer-direct delivery is mandatory unless 阿辉 explicitly asks for unified forwarding.
• For team-wide rule changes, prefer updating the shared skill/reference over copying edits into multiple places.

Agent Routing Reminder

• 小美 → news / information
• 大龙 → ops / diagnosis / system checks
• 元气 → creative / image / music / prompt output
• 小语 → coordination / review / routing / fallback only

If updating standards, keep all four aligned.