# smoke-monkey-harness-mcp

Source: https://glimind.com/server/npm%2Fsmoke-monkey-harness-mcp
Registry id: `npm/smoke-monkey-harness-mcp`
Status: unknown

## Install

```json
{
  "mcpServers": {
    "smoke-monkey-harness-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "smoke-monkey-harness-mcp"
      ]
    }
  }
}
```

Claude Code:

```sh
claude mcp add smoke-monkey-harness-mcp -- npx -y smoke-monkey-harness-mcp
```

## Tools (22)

- `harness_api`: Authoritative API reference (optionally sliced by area)
- `harness_events`: Complete agent event catalog for UI wiring
- `harness_examples`: List the bundled example agents
- `harness_guide`: Master instructions for building a looping agent (read first)
- `harness_guide_errors_validation_and_pauses`: Structured errors, tool-input validation, the three pauses
- `harness_guide_events_streaming_and_ui_wiring`: Event catalog + reference UI wiring
- `harness_guide_loop_phases_guards_and_compaction`: Loop phases, automatic guards, compaction, budgets
- `harness_guide_mcp_servers_and_discovery`: MCP server config, lazy activation, enable-vs-add approval flow
- `harness_guide_permissions_the_three_pauses`: The three pauses and how to resolve each
- `harness_guide_providers_models_and_api_keys`: Provider list, env keys, streaming, model selection
- `harness_guide_skills_skill_md_discovery`: SKILL.md format, discovery, list_skills/use_skill loading
- `harness_guide_storage_sessions_runs_messages`: Storage interface, sessions/runs/messages, resume
- `harness_guide_subcontexts_activation_and_switching`: On-demand guidance blocks, context_manage, built-in catalog
- `harness_guide_tools_custom_tool_implementation`: ToolDefinition shape, built-in factories, custom tools
- `harness_guide_ui_bridge_and_components`: Connecting @smoke-monkey/ui: help chat, widget, or full app
- `harness_plan`: Turn a product goal into a concrete build plan
- `harness_read_example`: Read one bundled example agent verbatim
- `harness_scaffold`: Generate a starter agent project on disk
- `harness_skill_content`: Load one bundled skill's full SKILL.md workflow
- `harness_skills_by_category`: Browse the 25 bundled agent-skills by category or domain
- `harness_status`: Installed library + server capabilities
- `harness_verify`: Typecheck/build an existing agent project

Measured by Glimind (https://glimind.com): MCP handshake + tool list only, no tool is ever called. CC BY 4.0.