Overview
Connect Vruum to an agent harness, learn MCP, skills, plugins, and CLI, and run revenue workflows from chat.
Vruum is operated from the AI harness you already use. The browser is the control panel for Dashboard, Docs, and Settings. The actual work — people, accounts, deals, queues, campaigns, research, marketing, content, tasks, team skills, and proposals — runs through MCP tools and Vruum skills.
The clean mental model:
- Plugin: the easiest install path when your harness supports one. It packages MCP connection details, skills, and instructions.
- MCP: the live tool connection your agent calls to read and write Vruum.
- Skills: repeatable operating workflows like pipeline fill, outreach triage, engagement triage, and deal review.
- CLI: technical fallback for local setup, CI, local files, and debugging. Non-technical clients should not need it for day-to-day use.
Start here
- Quick Start — copy the install prompt, choose plugin, direct MCP, or technical CLI setup, then verify Vruum.
- Skills — the operating workflows agents should invoke.
Core references
| Page | What it answers |
|---|---|
| Plugin | Which harnesses can install Vruum as a plugin, and what to use when plugin rails are unavailable. |
| MCP | How to connect the hosted MCP server, authenticate, verify setup, and when to use the local CLI. |
| Skills | How to install, discover, invoke, and publish skills, plus the common operating workflows. |
| CLI Reference | The technical fallback command tree, --json, and exit codes. |
Workflow skills
Skills automate the workflows. Each skill page also shows the underlying MCP and CLI calls so an agent can inspect, verify, or recover.
| Workflow | Use when |
|---|---|
| Pipeline fill | Source, research, gate, import, and enroll qualified prospects. |
| Outreach triage | Review pending email and LinkedIn outreach drafts. |
| Engagement triage | Review LinkedIn warming, nurture, marketing engagement, and content drafts. |
| Deal triage | Review the pipeline; surface at-risk deals and the next action for each. |
Agent-readable docs
Agents should not scrape the rendered browser page. Use the text routes. There
is one site-level llms.txt and one site-level llms-full.txt; individual
pages are exposed as .md.
| URL | Purpose |
|---|---|
https://vruum.ai/llms.txt | Small docs index for discovery. |
https://vruum.ai/llms-full.txt | Full public docs corpus in one response. |
https://vruum.ai/docs/<page>.md | Any single docs page as clean Markdown. |
https://vruum.ai/docs/skills/<skill>.md | Any skill page as clean Markdown. |
Product boundary
If you want a list, ask the agent for the list. If you want a row changed, ask the agent to change it. If you want an explanation, ask the agent to fetch the object and summarize the evidence.
The web app intentionally stays minimal: Dashboard, Docs, Settings, and proof artifacts. That keeps operating work inside the harness where the agent has context, tools, and a transcript.