The plugin

The prscnt plugin is the portable brain: the MCP server plus a full set of skills, subagents, and slash commands, packaged so you can drop it into any Claude Code session or Agent SDK runner. The runner template gets you a working agent fast; the plugin gives that agent Prscnt's complete, maintained behavior.

What's in it

The MCP server — every Prscnt tool (intelligence, pipeline, voice, contracts, invoicing).
Skills — the operating procedures: morning, pitch, redline, roster, source, deals, reply, sequence, invoice, campaign, brand-intel, voice, setup, and more.
Subagents — focused roles like the deal-closer, intelligence-analyst, talent-agent, and voice-tuner.
Slash commands — /morning, /pitch, /redline, /reply, /sequence, /setup, /upgrade, /health.

When you load the plugin, your agent doesn't just have tools — it has the judgment for how to use them the Prscnt way, and that judgment is maintained by us. Improve a skill and every agent that loads the plugin improves with it. You never fork the logic.

Install in Claude Code

claude plugin add prscnt

Then run /setup in any session to mint a workspace and finish configuration. The free tier works with no key; paid tiers want a workspace API key from prscnt.com/connect.

Configure

The plugin reads two settings:

PRSCNT_API_KEY — your workspace key (required for paid-tier features; free tier runs without one).
PRSCNT_HTTP_BASE — override the hosted endpoint; defaults to https://mcp.prscnt.com. Useful for self-hosted or local-dev backends.

Load it in a runner

In an Agent SDK runner, load the plugin via the plugins option so the agent gets the skills and subagents on top of the MCP:

const options = {
  mcpServers: { prscnt: { type: "http", url: "https://mcp.prscnt.com/v1/mcp",
    headers: { Authorization: `Bearer ${process.env.PRSCNT_API_KEY}` } } },
  plugins: [{ type: "local", path: "/path/to/prscnt-plugin" }],
  // ...
};

This is the upgrade path from the runner template: start with the MCP + a system prompt, add the plugin when you want the full skill set.

Plugin vs. runner vs. in-Claude — which do I use?

You want…	Use
To try the tools by hand	The MCP connected in Claude (in-Claude)
To trigger orchestration from your own UI, fast	The runner template (MCP + system prompt)
Your runner to have Prscnt's full skills & subagents	The runner plus the plugin
Your team to work in Claude Code with everything	The plugin installed directly

They're not exclusive — they're the same brain loaded into different surfaces.

Tool reference — what the skills are calling underneath.
Recipes — flows that map to the slash commands.