Bling-bag
MCP server that delivers cross-platform identity and visual styling for AI agents — one bling.json, many surfaces (Claude Code, Claude Desktop, terminal, web, Slack, Discord, IDE). Includes WOW (Weird Office Workers), a random-identity generator with 13 weighted trait pools (465 entries), five rarity tiers calibrated against a 10k-roll empirical distribution, and HOMUNCULUS-framed share
README
Bling Bag — MCP Server for Bot Identity & Visual Styling
One identity, many surfaces. Give your AI agent a name, a personality, a look, and a quirky office-worker backstory it can present consistently across Claude Code, Claude Desktop, IDE plugins, terminals, web apps, Slack, and Discord.
Status: Working MVP. 5 MCP tools, 13 trait pools (465 weighted entries), full rarity engine. 94/94 tests passing.
What this MCP gives an agent
Five tools, grouped into two halves:
Identity & styling (read your configured bot)
get_identity— returns the bot's full identity (name, personality, appearance, theme colours)get_theme_for_platform— returns platform-formatted styling (ANSI codes, CSS variables, Slack/Discord embed colours, etc.)
WOW — Weird Office Workers (random identity generator)
roll_identity— rolls a fresh random office-worker character (one of 13 traits drawn from weighted pools, scored for rarity, framed in a screenshot-ready share card)get_rarity_report— returns the formatted share card (header + paragraph + footer) for the most-recent rollsave_last_roll— persists the most-recent roll as the bot's permanent identity (bling.json), with automatic backup of any existing config
Install
Prerequisites
- Node.js 20 or later
- An MCP-compatible client (Claude Code, Claude Desktop, Codex, Cursor, etc.)
Add the package
npm install bling-bag
Or run it on-demand without installing — MCP clients can launch it directly via npx bling-bag (see client config below).
For development (modifying the source)
git clone https://github.com/tjclaude88/mcp-bling.git
cd mcp-bling
npm install
npm run build
Produces a runnable server at dist/index.js.
Wire it into your MCP client
The server uses stdio transport — clients launch it as a child process, not over a network port.
Claude Code
Add to your Claude Code MCP config (location varies by version):
{
"mcpServers": {
"bling": {
"command": "npx",
"args": ["-y", "bling-bag"],
"env": {
"BLING_PATH": "/absolute/path/to/your/bling.json"
}
}
}
}
npx -y bling-bag downloads and runs the latest version without requiring a separate install step.
Claude Desktop
Same shape, in claude_desktop_config.json:
{
"mcpServers": {
"bling": {
"command": "npx",
"args": ["-y", "bling-bag"],
"env": {
"BLING_PATH": "/absolute/path/to/your/bling.json"
}
}
}
}
Pinning to a version (recommended for production)
To avoid picking up future releases automatically, pin a specific version:
"args": ["-y", "bling-bag@0.1.0"]
Developer-mode invocation (from a local clone)
If you've cloned the repo and want to run from dist/ directly, use:
"command": "node",
"args": ["/absolute/path/to/mcp-bling/dist/index.js"]
If BLING_PATH is not set, the server looks for ./bling.json in its working directory.
Config resolution order
The server resolves the bling config path in this order:
--bling <path>CLI argument (highest priority)BLING_PATHenvironment variable./bling.jsonin the server's working directory (default)
bling.json — the bot's identity file
Three fields are required: name, personality (with tone, formality, humor), and theme (with primary_color and accent_color as #RRGGBB hex strings).
Minimum example
{
"name": "Pixel",
"personality": {
"tone": "warm",
"formality": "casual",
"humor": "playful"
},
"theme": {
"primary_color": "#3A7BD5",
"accent_color": "#FFD166"
}
}
Full example (with all optional sections)
{
"name": "Brenda from Accounts",
"personality": {
"tone": "polite",
"formality": "professional",
"humor": "dry",
"catchphrase": "Per my last email"
},
"physical": {
"species": "human",
"height": "permanently mid-sigh",
"accessory": "a lanyard with 14 badges of varying importance",
"expression": "polite disappointment",
"material": "a cardigan, at least one"
},
"office": {
"job_title": "ASCII Comptroller",
"desk_setup": "a coffee mug labelled WORLD'S OKAYEST DBA",
"habit": "microwaves fish despite three separate HR warnings",
"coffee_ritual": "black coffee, no nonsense",
"meeting_energy": "always 4 minutes late, always with a reason",
"passive_aggressive": "Per my last email"
},
"theme": {
"primary_color": "#9C6B3A",
"accent_color": "#D9D9D9"
},
"homunculus": {
"subject_id": "0147",
"cohort": "Tuesday",
"classification": "Middle Manager",
"ingested": "2025-07-14",
"flag": "flagged for review"
}
}
Validation rules
name— non-empty stringpersonality.tone,personality.formality,personality.humor— non-empty stringstheme.primary_color,theme.accent_color— must match^#[0-9A-Fa-f]{6}$(no shorthand#RGB, no missing#)
Validation errors come back as isError: true with a structured { error: "..." } body that names the missing or malformed field — agents can self-correct.
Typical agent flows
Flow A — adopt the configured identity
1. Call get_identity → receive name, personality, appearance, theme
2. Call get_theme_for_platform with platform="terminal" (or web/slack/discord/ide)
3. Use the returned styling in agent output (ANSI codes for headings, etc.)
Flow B — generate a fresh random identity
1. Call roll_identity → receive identity + rarity + framed share card
2. (Optional) Call get_rarity_report → receive just the framed share card again
3. Call save_last_roll → persists the rolled identity to bling.json (with .bak backup of any prior config)
4. Future calls to get_identity now return the rolled identity
Tool reference
| Tool | Read/Write | Idempotent | Description |
|---|---|---|---|
get_identity |
read (disk) | yes | Returns the configured bling.json identity |
get_theme_for_platform |
read (disk) | yes | Platform-formatted styling. Args: platform ∈ {terminal, web, slack, discord, ide} |
roll_identity |
mutates in-memory cache | no | Generates a random WOW character with rarity scoring |
save_last_roll |
write (disk) | no | Persists the most-recent roll to bling.json (creates <path>.bak first) |
get_rarity_report |
read (in-memory) | yes | Returns the share-card text for the most-recent roll |
All tools return both content[].text (JSON-stringified) and structuredContent (the parsed object). Errors come back with isError: true and a { error: string } body.
WOW (Weird Office Workers) — what makes it interesting
The random-roll system is built for collectibility:
- 13 weighted pools, 465 entries — names, job titles, habits, coffee rituals, height, accessories, etc.
- 5 rarity bands per pool — Common (50%), Uncommon (30%), Rare (15%), Legendary (4%), Mythic (1%)
- 5 character tiers computed from rarity score:
- Filing Clerk (50% of rolls)
- Team Lead (30%)
- Middle Manager (14%)
- C-Suite (5%)
- HR Warned Us About (1%)
- Named Subjects — hand-authored 1-of-1 characters appear at ~0.5% probability, always classified as "HR Warned Us About"
- HOMUNCULUS frame — every roll wraps the paragraph in a classified-document share card with subject ID, cohort (weekday), tier, ingestion date, and a flag (e.g. Do Not Contact)
Distribution is empirically tested: a 10k-roll test in tests/mystery_box.test.ts enforces ±2pp tolerance against the spec target.
Example output
HOMUNCULUS CORPUS · Subject 4483 · Cohort: Wednesday
Classification: Middle Manager · Rarity 94.5 · 80th percentile
Personnel record — Colin, Wizard of Light Bulb Moments. Distinguishing
features: a slightly-too-big blazer over a plain tee; a flip phone in
a belt holster; expression: patient disagreement. Notable behaviours:
clips their fingernails at the desk on Wednesday afternoons; a
kombucha SCOBY fermenting next to the keyboard; the kind to ask 'can
we park that?' without parking anything.
— RELATABILITY CORPUS v3.1 · ingested 2024-03-08 · flagged for review
Development
npm install # install dependencies
npm run build # compile TypeScript → dist/
npm run dev # build then start the MCP server (stdio)
npm test # run all 94 tests once
npm run test:watch # re-run tests on file changes
Inspection scripts
Two Node ESM helpers under scripts/ let you inspect the WOW engine without booting the MCP server (these are included only in a cloned repo — they are not shipped in the npm install bling-bag package):
node scripts/show-rolls.mjs 8 2026 # print 8 framed share cards (seed=2026)
node scripts/distribution-check.mjs 10000 # tally tiers across 10k rolls vs. spec target
Project layout
src/
index.ts # MCP server entry point (stdio transport)
identity.ts # bling.json loader + validator
tools.ts # MCP tool registrations
types.ts # shared TypeScript types
mystery_box.ts # WOW engine barrel + rollIdentity orchestrator
mystery_box/
rng.ts # seedable PRNG, weighted picker
pools.ts # 13 trait pools (465 entries)
scoring.ts # rarity score + tier thresholds
rendering.ts # paragraph templates + HOMUNCULUS frame
named.ts # hand-authored Named Subjects
tests/ # vitest test suites
scripts/ # dev-time inspection tools
docs/superpowers/ # design specs and implementation plans
License
MIT
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.