Skip to main content
Glama
RaedBerjaoui

Vibecoders MCP

by RaedBerjaoui

Server Configuration

Describes the environment variables required to run the server.

NameRequiredDescriptionDefault
GITHUB_TOKENNoGitHub token (optional)
VERCEL_TOKENNoVercel token (optional)
BRAVE_API_KEYNoBrave API key (optional)
GEMINI_API_KEYNoGemini API key (optional)
OPENAI_API_KEYNoOpenAI API key (optional)
TAVILY_API_KEYNoTavily API key (optional)
ANTHROPIC_API_KEYNoAnthropic API key (optional)
VIBECODERS_SERVERSNoPath to a custom servers.json file (resolved in order: this env var, then ~/.vibecoders/servers.json, then ./servers.json)
SUPABASE_ACCESS_TOKENNoSupabase access token (optional)

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

CapabilityDetails
tools
{
  "listChanged": true
}

Tools

Functions exposed to the LLM to take actions

NameDescription
doctorA

Report Vibecoders status: capabilities, delegation providers, mounted servers, optional API keys, tool groups, and this session’s lane. Pass json:true for structured output. Never prints secret values.

list_providersA

List delegation providers whose CLI is installed and how each bills. Delegation routes a task to another coding agent on your subscription (its CLI), not a metered API.

delegateA

Hand a task to another coding agent via its CLI — bills through your subscription, not an API key. Read-only by default; pass mode:"write" to let it modify files under cwd. Pass background:true to run it as a non-blocking task you manage with tasks_list / tasks_steer / tasks_interrupt. Call list_providers first to see what is installed.

generate_imageA

Generate an image from a text prompt and save it to an absolute file path. Uses your configured image_gen provider (codex CLI on your ChatGPT plan — no API key — or OpenAI/Gemini with your own key). If none is set up, it tells you how to enable one. Pass an absolute out_path; the saved file path is returned.

web_searchA

Search the web for a query and return ranked results (and a synthesized answer when the provider offers one). Uses your configured web_search provider (gemini CLI on your Google plan — no API key — or Gemini/Brave/Tavily with your own key). If none is set up, it tells you how to enable one.

search_toolsA

Search across every mounted MCP server for tools matching a query. Returns qualified ids (server.tool); load one with load_tool before calling it.

load_toolA

Get the full input schema for a qualified tool id (server.tool), so you can call it correctly.

call_toolC

Call a qualified tool (server.tool) on a mounted MCP server with arguments.

write_handoffA

Save a handoff for THIS session's lane (app@detached) so a future session can resume exactly here. Lanes are isolated — concurrent sessions never mix handoffs.

recall_handoffA

Load the latest handoff for THIS session's lane (app@detached), if any.

memory_storeA

Save a durable memory to the graph (RAG-recallable later). Use for decisions, facts, gotchas, and links between them. scope: "project" (this repo/branch, the default), "global" (all your work), or a custom bucket name. Pass a stable id and links to build a graph you can walk.

memory_recallA

Recall memories by meaning (RAG). Searches global + this project's memory by default. scope: "all" (default), "project", "global", or a custom bucket.

memory_walkA

Walk the memory graph from a node along its links (the connected context around a decision). direction: "out" (default), "in", or "both".

memory_forgetA

Forget a memory by id (appends a tombstone; it stops appearing in recall/walk).

project_contextA

Orient in THIS project fast: branch, recent commits, the saved handoff, and the project's most recent memories — one call, no git subprocess.

reference_inspectA

Study a web page's technique: title, description, detected frameworks/libraries, headings, and top links. Fetches the live page (SSRF-guarded, size-capped). Use to understand how a reference is built — not to copy it.

reference_excerptA

Fetch a page and return a focused plain-text excerpt — around query if given, otherwise the head. SSRF-guarded and capped. For reading a specific passage.

reference_read_sourceA

Fetch and return the FULL source of a page (HTML/text), SSRF-guarded and capped. The escape hatch when inspect/excerpt are not enough — requires a reason.

tasks_listA

List background delegations (those started with delegate background:true), newest first. Pass an id to get that task’s full output.

tasks_steerA

Send a line of input to a running background task’s stdin. Best-effort: one-shot delegate CLIs may not accept input mid-run (the tool will say so).

tasks_interruptA

Stop a running background task (SIGTERM) and mark it interrupted.

design_coreA

Load the design-intelligence core of the anti-AI-design RAG: the standard that makes the output NOT read as AI-made and hold its formatting across every screen, the tells principle (displace the bias, never the surface instance), the build non-negotiables, and the menu of deeper layers. Call this FIRST when building ANY UI, website, page, component, or visual, then pull a layer with design_layer. Opt-in (features.rag).

design_layerA

Pull ONE deeper design layer on demand, by name: donts (the vibecoded-tells catalogue — self-check the plan against it before shipping), formatting (the laws that make a page fill any screen with no dead margin and nothing clipped), directives (hard build directives: how to source visuals and structure a one-page site), scaffolds (occupancy-correct section scaffolds and their CSS — one dominant each), type_pointers (formatting-safe typographic elevation), image_gen (how to generate imagery — render it with the generate_image tool). Call design_core first; pull a layer only when the build needs it. Opt-in (features.rag).

Prompts

Interactive templates invoked by user choice

NameDescription

No prompts

Resources

Contextual data attached and managed by the client

NameDescription

No resources

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/RaedBerjaoui/vibecoders-mcp'

If you have feedback or need assistance with the MCP directory API, please join our Discord server