Fiber AI Plugin

Use Fiber AI with any AI coding agent. Search companies, find contacts, enrich data, and manage audiences — directly from Claude Code, Cursor, OpenCode, Gemini CLI, Codex CLI, GitHub Copilot CLI, VS Code, Windsurf, and 40+ other AI agents.

Fiber MCP ships three HTTP endpoints:

All three use Streamable HTTP transport. You can register any combination; the MCP server names stay distinct (fiber-ai-v2, fiber-ai-v3, fiber-ai-core).

All install paths give you the MCP servers, 16 built-in skills (8 general + 8 playbook workflows), and 7 persona subagents (ai-recruiter, ai-sdr, gtm-strategist, signal-scout, data-quality-auditor, product-engineer, fiber-sde) that do the work like a human teammate with full Fiber product knowledge.

/plugin marketplace add fiber-ai/fiber-ai-plugin
/plugin install fiber@fiber-tools

This installs MCP tools, skills, rules, and hooks in one shot. Run /fiber:help to list commands.

Team sharing (optional) — commit a .mcp.json at the project root so teammates get Fiber AI automatically:

{
  "mcpServers": {
    "fiber-ai-v2": { "type": "http", "url": "https://mcp.fiber.ai/mcp/v2" },
    "fiber-ai-v3": { "type": "http", "url": "https://mcp.fiber.ai/mcp/v3" },
    "fiber-ai-core": { "type": "http", "url": "https://mcp.fiber.ai/mcp" }
  }
}

Option A — Deeplink (paste in your browser, Cursor must be installed):

V2 (recommended for most users, API key auth):

cursor://anysphere.cursor-deeplink/mcp/install?name=fiber-ai-v2&config=eyJ0eXBlIjoiaHR0cCIsInVybCI6Imh0dHBzOi8vbWNwLmZpYmVyLmFpL21jcC92MiJ9

V3 (every public operation, OAuth / SSO auth):

cursor://anysphere.cursor-deeplink/mcp/install?name=fiber-ai-v3&config=eyJ0eXBlIjoiaHR0cCIsInVybCI6Imh0dHBzOi8vbWNwLmZpYmVyLmFpL21jcC92MyJ9

Core (5 meta-tools over every public operation):

cursor://anysphere.cursor-deeplink/mcp/install?name=fiber-ai-core&config=eyJ0eXBlIjoiaHR0cCIsInVybCI6Imh0dHBzOi8vbWNwLmZpYmVyLmFpL21jcCJ9

Option B — Manual: copy cursor/mcp.json from this repo to your project's .cursor/mcp.json. Optionally copy cursor/rules/fiber-api.mdc to .cursor/rules/ for inline agent guidance.

Persona subagents: Cursor picks up subagents from .cursor/agents/. Copy the persona files into your project:

mkdir -p .cursor/agents && cp path/to/fiber-ai-plugin/cursor/agents/*.md .cursor/agents/

In your OpenCode session, tell the agent:

Fetch and follow instructions from https://raw.githubusercontent.com/fiber-ai/fiber-ai-plugin/main/.opencode/INSTALL.md

The agent will register the Fiber MCP servers (V2, V3, Core), clone the skills, copy the persona subagents into .opencode/agents/, and append a "Fiber AI skills and personas" section to your project AGENTS.md.

gemini extensions install https://github.com/fiber-ai/fiber-ai-plugin

To update later:

gemini extensions update fiber-ai

The repo ships a gemini-extension.json that registers all three Fiber MCP endpoints (V2, V3, Core) and exposes the skills via the AGENTS.md context file.

Codex CLI reads MCP servers from ~/.codex/config.toml. Append:

[mcp_servers.fiber-ai-v2]
url = "https://mcp.fiber.ai/mcp/v2"
transport = "http"

[mcp_servers.fiber-ai-v3]
url = "https://mcp.fiber.ai/mcp/v3"
transport = "http"

[mcp_servers.fiber-ai-core]
url = "https://mcp.fiber.ai/mcp"
transport = "http"

Then export your API key (V2 and Core support API-key auth; V3 uses OAuth SSO instead):

export FIBER_API_KEY=sk_live_...

To load the playbook skills, clone this repo next to your project and reference it in your Codex system prompt, or copy individual skills/*/SKILL.md files into your Codex skills directory.

Persona subagents for Codex CLI: the plugin ships TOML-format subagents under .codex/agents/. Copy them into your Codex agents directory:

mkdir -p .codex/agents && cp path/to/fiber-ai-plugin/.codex/agents/*.toml .codex/agents/

Then invoke any of @ai-recruiter, @ai-sdr, @gtm-strategist, @signal-scout, @data-quality-auditor, @product-engineer, or @fiber-sde inside Codex.

Copilot CLI reads MCP servers from ~/.config/github-copilot/mcp.json:

{
  "mcpServers": {
    "fiber-ai-v2": { "type": "http", "url": "https://mcp.fiber.ai/mcp/v2" },
    "fiber-ai-v3": { "type": "http", "url": "https://mcp.fiber.ai/mcp/v3" },
    "fiber-ai-core": { "type": "http", "url": "https://mcp.fiber.ai/mcp" }
  }
}

Copilot CLI picks up AGENTS.md automatically — clone or symlink this repo into your workspace to surface the skills to the agent.

Add to your .vscode/mcp.json:

{
  "mcpServers": {
    "fiber-ai-v2": { "type": "http", "url": "https://mcp.fiber.ai/mcp/v2" },
    "fiber-ai-v3": { "type": "http", "url": "https://mcp.fiber.ai/mcp/v3" },
    "fiber-ai-core": { "type": "http", "url": "https://mcp.fiber.ai/mcp" }
  }
}

The skills live in skills/ — VS Code agents such as GitHub Copilot Chat honor the AGENTS.md file in your workspace root.

Add the three MCP servers in Windsurf Settings (Transport: HTTP):

• V2 (API key): https://mcp.fiber.ai/mcp/v2
• V3 (OAuth): https://mcp.fiber.ai/mcp/v3
• Core (API key): https://mcp.fiber.ai/mcp

Copy windsurf/rules/fiber-api.md from this repo to your Windsurf rules directory.

Install individual skills for any agent that supports skills.sh:

npx skills add fiber-ai/fiber-ai-plugin --skill find-similar-companies
npx skills add fiber-ai/fiber-ai-plugin --skill enrich-linkedin-csv
npx skills add fiber-ai/fiber-ai-plugin --skill build-recruiting-audience
npx skills add fiber-ai/fiber-ai-plugin --skill expand-from-email-list
npx skills add fiber-ai/fiber-ai-plugin --skill enrich-github-handles
npx skills add fiber-ai/fiber-ai-plugin --skill find-and-enrich-by-role
npx skills add fiber-ai/fiber-ai-plugin --skill track-signals
npx skills add fiber-ai/fiber-ai-plugin --skill benchmark-vs-competitor
npx skills add fiber-ai/fiber-ai-plugin --skill search
npx skills add fiber-ai/fiber-ai-plugin --skill enrich
npx skills add fiber-ai/fiber-ai-plugin --skill audience
npx skills add fiber-ai/fiber-ai-plugin --skill sdk-ts
npx skills add fiber-ai/fiber-ai-plugin --skill sdk-py
npx skills add fiber-ai/fiber-ai-plugin --skill quickstart
npx skills add fiber-ai/fiber-ai-plugin --skill setup
npx skills add fiber-ai/fiber-ai-plugin --skill help

1. Get your API key from fiber.ai/app/api.

2. Set it as an environment variable:

   export FIBER_API_KEY=sk_live_...

   Add to ~/.zshrc / ~/.bashrc for persistence.

The MCP servers read your key from the client's request headers. Nothing is stored inside the plugin itself.

Fiber ships 7 persona subagents + 16 skills. Personas are the "hire an AI teammate" unit - they carry domain expertise (recruiting, sales, GTM, signal tracking, data QA, product integration, software development) plus full Fiber product knowledge so the user does not have to teach their LLM about either. Skills are the underlying workflow primitives the personas call into.

Each persona:

• Comes preloaded with domain expertise (recruiter tradeoffs vs SDR tradeoffs vs GTM strategy vs signal noise filtering vs benchmarking rigor).
• Knows every relevant Fiber operationId by name (no meta-tool round-trips for common operations).
• Calls into the plugin's skills automatically rather than hand-rolling HTTP calls.
• Enforces the same cost / consent gates as the skills - never charges silently.
• @data-quality-auditor never recommends a vendor - it presents numbers (including where Fiber underperformed) and lets the user decide.

For building applications programmatically:

• TypeScript: npm install @fiberai/sdk — GitHub
• Python: pip install fiberai — GitHub

Both SDKs ship their own llms.txt pointing back at the canonical API docs.

For open-source examples, reference implementations, and community patterns: github.com/fiber-ai/open-fiber.

Point your LLM / coding agent at these endpoints when building against Fiber:

• Routing index + critical rules: https://api.fiber.ai/llms.txt
• Operation index (Stripe-style): https://api.fiber.ai/ai-docs/index.md
• Per-operation pages: https://api.fiber.ai/ai-docs/<operationId>.md (e.g. companySearch, syncQuickContactReveal, syncTurboContactEnrichment, triggerExhaustiveContactEnrichment)
• Full concatenated corpus (RAG): https://api.fiber.ai/llms-full.txt
• OpenAPI (JSON): https://api.fiber.ai/openapi.json — send Accept: text/markdown on the same URL for the agent-friendly markdown index
• MCP: https://mcp.fiber.ai/mcp/v2 (direct tools, API key) · https://mcp.fiber.ai/mcp/v3 (direct tools over every public operation, OAuth / SSO) · https://mcp.fiber.ai/mcp (5 meta-tools, API key)
• MCP quickstart: https://docs.fiber.ai/article/using-mcp-in-llms

Authoring rules for anyone extending this plugin: see AGENTS.md.

• Fiber AI
• Fiber Documentation
• API Documentation
• API Key Management
• Credits and Billing

MIT — Fiber AI 2026