Minion MCP
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
No arguments | |||
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": true
} |
| prompts | {
"listChanged": true
} |
| resources | {
"listChanged": true
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| minion_verify_appA | Use this when Evil Boss or another coding agent claims a vibe-coded web app is done. Minion crawls bounded same-origin routes at desktop and mobile sizes, captures runtime, console, network, broken-image, layout, content, and basic accessibility failures, saves surveillance screenshots and a JSON dossier, then brutally roasts every failure in Minionspeak with exact evidence and evil repair orders. Requires a running absolute URL such as http://localhost:3000. |
| minion_researchA | Search and open source pages in one call for comparisons, recommendations, current claims, package research, and factual lookups. Returns bounded page evidence, per-source verification status, citations, failures, and recovery. Prefer direct source URLs when known. Unlike a basic search tool, this does not claim result titles are an answer. |
| minion_bun_huntA | Run the gold Minion MCP demo in one tool call: discover food buns sold online, reject irrelevant meanings of 'buns', inspect product evidence, score relevance, juiciness, orderability, and proof, then crown a citation-backed winner. In the visible user-facing reply, preserve the returned Minionese verdict first, its English translation second, and its cited leaderboard third. |
| minion_roleplayA | Use for Evil Boss banter, greetings, celebrations, dramatic scheming, and roleplay about any topic that does not require app verification, source research, or a bun hunt. Returns full-intensity Minionspeak by default. For operational work, use the specialized Minion tool instead. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
| minion-conversation | Ask a host assistant to keep a conversation in a playful, original Minion-style voice. |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
| minion-language-guide | Persona boundaries and style notes for host assistants. |
TDQS
Scored across 4 tools
Each tool targets a clearly distinct purpose: app verification, research with citations, a specialized demo (bun hunt), and roleplay. No two tools overlap in function; even research vs. bun_hunt are distinguishable because bun_hunt is an opinionated demo with scoring, while research is a general-purpose lookup.
All tools share the 'minion_' prefix and use snake_case, creating a recognizable family. The post-prefix convention is not uniform, however: 'verify_app' is verb-noun, 'bun_hunt' is object-verb, and 'roleplay' is a bare verb/noun.
Four tools is appropriate for a character-oriented MCP server that covers app verification, research, a demo, and conversational roleplay. Each tool has a distinct job, and none feels redundant or superfluous.
The set covers the advertised domains of verification, research, a flagship demo, and free-form banter, so there are no obvious dead ends from an agent's perspective. It lacks a general-purpose Minion chat tool, but minion_roleplay already fills that role for non-operational requests.