The Spotter.

What this is

A second pair of eyes that lifts the work.

A spotter in powerlifting watches your form, catches the bar if something breaks down, and gives you the confidence to attempt a lift you couldn't safely attempt alone. They lift you, not the bar.

This skill does the same for an epic. It walks the work through nine product-leadership lenses — empathy, competitive landscape, strategic differentiation, solution approach, holistic impact, packaging, launch readiness, post-launch ownership, and trust/governance/auditability — and produces structured feedback in the critique-not-criticism tradition. Every flag is "you could strengthen this by..." Never "you missed."

The PM still owns the lift. The Spotter is just there to lift them.

See what a Spotter review looks like →

Connect

Plug The Spotter into your agent.

Claude Code

claude mcp add spotter https://spotter-mcp.missionbuilt.io/sse

Claude Desktop

Add to claude_desktop_config.json:

{
  "mcpServers": {
    "spotter": {
      "url": "https://spotter-mcp.missionbuilt.io/sse"
    }
  }
}

Other MCP clients

• https://spotter-mcp.missionbuilt.io/sseServer-Sent Events transport (most common)
• https://spotter-mcp.missionbuilt.io/mcpStreamable HTTP transport (newer)

Once connected, ask your agent: "run the spotter on this epic" and paste an epic. The framework loads from this server; your agent does the reasoning.

Tools exposed

Seven tools, three resources.

• spotter_get_skillReturns the full SKILL.md framework
• spotter_get_examplesReturns 64 worked lens examples
• spotter_get_calibration_epicReturns the synthetic test epic
• spotter_list_lensesReturns the nine lenses with weights
• spotter_get_brand_cssReturns the Mission Built design CSS
• spotter_reviewPrimes review mode on a pasted epic
• spotter_buildPrimes build mode for a new epic
• spotter_iteratePrimes iterate mode on a draft

Open source

Take it. Fork it. Lift others with it.

The Spotter ships under MIT. Use it commercially, adapt it for your team, embed it in your own product. The skill markdown is in The Loadout repo on GitHub alongside the source for this MCP server. Full attribution to the open-source skills and product writers whose thinking informed it lives in spotter/ATTRIBUTION.md.

Customize for your team by adding a CLAUDE.md at your project root with your tier names, competitor set, and product taxonomy. The skill stays generic; your context stays local.