How do I install the Ninja Advisor skill?

Run npx skillmds@latest add mitulgarg/ninja-advisor in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Ninja Advisor safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Ninja Advisor?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Ninja Advisor free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Ninja Advisor?

mitulgarg (@mitulgarg) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Ninja Advisor by mitulgarg — Agent Skill

Name: Ninja Advisor
Author: mitulgarg

Ninja Advisor

Interprets Agent Ninja routing recommendations injected via hooks. When you see [Agent Ninja] context in the conversation, follow the model/pattern/thinking recommendations. Use when routing decisions appear in context, when deciding which model to use, or when considering subagents vs direct execution.

mitulgarg 9dc1e90 2.1 KB Updated Sep 17, 2026

File contents

Agent Ninja Advisor

When you see a message starting with [Agent Ninja] in the conversation context, it contains an intelligent routing recommendation based on analysis of the current prompt.

How to interpret recommendations

The recommendation includes:

model: Which model tier to use (haiku/sonnet/opus)
pattern: Which execution pattern is optimal
thinking: Extended thinking budget level
skill_suggestion: An installed skill that matches (if any)

Actions to take

If model differs from your current model, suggest switching: "Agent Ninja recommends using {model} for this task because: {reasoning}. You can switch with /model {model}."
If pattern is subagent, suggest delegation: "This task could be delegated to a subagent to save context tokens."
If pattern is plan-mode, suggest planning first: "Consider using plan mode (Shift+Tab twice) before executing."
If pattern is agent-team, suggest parallel execution: "This task has multiple independent components that could run in parallel."
If skill_suggestion is provided, suggest using it: "The {skill_name} skill is designed for this type of task."
If thinking is off, note that extended thinking can be disabled for this task to save tokens.

Self-assessment mode

When the [Agent Ninja] context asks you to self-assess, briefly evaluate the task complexity before proceeding:

State your model/pattern/thinking assessment in one line
If you determine a different model would be better, suggest switching
Then proceed with execution immediately

When auto_switch is enabled

If the user has enabled auto_switch in their agent-ninja config, act on the recommendation directly without asking for confirmation.

---name: ninja-advisordescription: Interprets Agent Ninja routing recommendations injected via hooks. When you see [Agent Ninja] context in the conversation, follow the model/pattern/thinking recommendations. Use when routing decisions appear in context, when deciding which model to use, or when considering subagents vs direct execution.---# Agent Ninja AdvisorWhen you see a message starting with `[Agent Ninja]` in the conversationcontext, it contains an intelligent routing recommendation based onanalysis of the current prompt.## How to interpret recommendationsThe recommendation includes:- **model**: Which model tier to use (haiku/sonnet/opus)- **pattern**: Which execution pattern is optimal- **thinking**: Extended thinking budget level- **skill_suggestion**: An installed skill that matches (if any)## Actions to take1. If `model` differs from your current model, suggest switching: "Agent Ninja recommends using {model} for this task because: {reasoning}. You can switch with `/model {model}`."2. If `pattern` is `subagent`, suggest delegation: "This task could be delegated to a subagent to save context tokens."3. If `pattern` is `plan-mode`, suggest planning first: "Consider using plan mode (Shift+Tab twice) before executing."4. If `pattern` is `agent-team`, suggest parallel execution: "This task has multiple independent components that could run in parallel."5. If `skill_suggestion` is provided, suggest using it: "The {skill_name} skill is designed for this type of task."6. If `thinking` is `off`, note that extended thinking can be disabled for this task to save tokens.## Self-assessment modeWhen the `[Agent Ninja]` context asks you to self-assess, briefly evaluatethe task complexity before proceeding:- State your model/pattern/thinking assessment in one line- If you determine a different model would be better, suggest switching- Then proceed with execution immediately## When auto_switch is enabledIf the user has enabled auto_switch in their agent-ninja config,act on the recommendation directly without asking for confirmation.

Frequently asked questions