ui-automation-control-plane

How-To: MCP Client Setup

Prerequisites

• Run from repo root.
• pnpm install completed.
• MCP-capable client enabled.

This setup works with MCP-capable clients such as Codex, Claude Code, or any other MCP host that can launch a stdio server. That is a protocol compatibility statement, not a partnership claim.

Installation Truth

• Publish-ready package target: @uiq/mcp-server
• Planned CLI / bin name: prooflane-mcp
• Protocol today: stdio only
• Auth boundary today: local stdio startup does not use OAuth; protected HTTP/API and automation surfaces keep the existing token/header contract
• The truthful package-launch shape is npx -y @uiq/mcp-server / pnpm dlx @uiq/mcp-server, and it is not published yet

Default Behavior: Core 12

By default the server exposes only the 12 core tools below. No extra env is required:

1. uiq_backend_runtime
2. uiq_api_sessions
3. uiq_api_flows
4. uiq_api_templates
5. uiq_api_runs
6. uiq_catalog
7. uiq_server_selfcheck
8. uiq_run_profile
9. uiq_run_stream
10. uiq_run_overview
11. uiq_read_artifact
12. uiq_gate_failures

Enable Optional Tool Groups

Use UIQ_MCP_TOOL_GROUPS to enable optional groups on demand: advanced/register/proof/analysis.

# Example: enable only register + proof
UIQ_MCP_TOOL_GROUPS=register,proof

# Example: enable every optional group
UIQ_MCP_TOOL_GROUPS=all

Legacy compatibility switches are still accepted:

• UIQ_MCP_EXPOSE_ADVANCED_TOOLS=true
• UIQ_MCP_ENABLE_ADVANCED_TOOLS=true

Either legacy switch enables every optional group.

MCP Config Template

Repo-native today

{
  "mcpServers": {
    "uiq": {
      "command": "pnpm",
      "args": ["mcp:start"],
      "cwd": "/ABSOLUTE/PATH/TO/REPO",
      "env": {
        "UIQ_MCP_API_BASE_URL": "http://127.0.0.1:18080",
        "UIQ_MCP_TOOL_GROUPS": "advanced,analysis,proof"
      }
    }
  }
}

Publish-ready package shape (not live yet)

{
  "mcpServers": {
    "uiq": {
      "command": "npx",
      "args": ["-y", "@uiq/mcp-server"],
      "env": {
        "UIQ_MCP_API_BASE_URL": "http://127.0.0.1:18080",
        "UIQ_MCP_TOOL_GROUPS": "advanced,analysis,proof"
      }
    }
  }
}

That second config shape is the future package path for @uiq/mcp-server. It is ready to document, but it is not published yet.

Full Registered Tool Inventory (40)

1. uiq_a11y_top
2. uiq_api_automation_cancel
3. uiq_api_automation_commands
4. uiq_api_automation_run
5. uiq_api_automation_task
6. uiq_api_automation_tasks
7. uiq_api_flows
8. uiq_api_runs
9. uiq_api_sessions
10. uiq_api_templates
11. uiq_backend_runtime
12. uiq_catalog
13. uiq_compare_perf
14. uiq_computer_use_run
15. uiq_diff_proof_campaign
16. uiq_export_proof_bundle
17. uiq_gate_failures
18. uiq_list_runs
19. uiq_model_target_capabilities
20. uiq_perf_metrics
21. uiq_read_artifact
22. uiq_read_manifest
23. uiq_read_proof_report
24. uiq_read_repo_doc
25. uiq_register_orchestrate
26. uiq_register_state
27. uiq_run_command
28. uiq_run_overview
29. uiq_run_profile
30. uiq_run_proof_campaign
31. uiq_run_stream
32. uiq_security_summary
33. uiq_server_selfcheck
34. uiq_summarize_failures
35. uiq_visual_status
36. uiq_read_run_ai_review
37. uiq_generate_release_brief
38. uiq_find_similar_failures
39. uiq_explain_template_feasibility
40. uiq_list_manual_gates

Role-Based Starting Paths

Operator

Use this path when you need to run work, inspect the current manual-gate inbox, and read a release summary without inventing a second truth surface.

1. uiq_server_selfcheck
2. uiq_run_profile or uiq_run_stream
3. uiq_run_overview
4. uiq_list_manual_gates
5. uiq_generate_release_brief

Agent / MCP Client

Use this path when an agent needs governed read context before proposing the next action.

1. uiq_catalog
2. uiq_read_run_ai_review
3. uiq_generate_release_brief
4. uiq_find_similar_failures
5. uiq_explain_template_feasibility

Recommended MCP Review Loop

Use this loop when a Codex session, Claude Code session, or another MCP-capable client needs to understand the latest result without turning Prooflane into a generic chat tool:

1. uiq_run_overview
2. uiq_generate_release_brief
3. uiq_find_similar_failures
4. uiq_list_manual_gates
5. uiq_explain_template_feasibility

Why this loop is truthful:

• uiq_generate_release_brief is a read-only summary over governed evidence
• uiq_find_similar_failures reuses historical failure artifacts instead of inventing a second memory system
• uiq_list_manual_gates and uiq://manual-gates/inbox-summary expose the same paused-run context shown in Runs & Blocks
• uiq_explain_template_feasibility exposes the same migration hints that the UI now surfaces in Advanced Review

Release Reviewer

Use this path when you want to compare existing governed evidence and inspect why a candidate is blocked, ready, or still missing proof.

1. uiq_run_proof_campaign
2. uiq_read_proof_report
3. uiq_diff_proof_campaign
4. uiq_generate_release_brief
5. uiq_compare_perf

Suggested Operating Sequence

1. uiq_server_selfcheck
2. uiq_catalog
3. uiq_run_profile or uiq_run_stream
4. uiq_run_overview
5. uiq_gate_failures (if not passed)
6. uiq_read_artifact

run override fields accepted (runOverrideSchema)

• baseUrl
• app
• bundleId
• diagnosticsMaxItems
• exploreBudgetSeconds
• exploreMaxDepth
• exploreMaxStates
• chaosSeed
• chaosBudgetSeconds
• chaosClickRatio
• chaosInputRatio
• chaosScrollRatio
• chaosKeyboardRatio
• loadVus
• loadDurationSeconds
• loadRequestTimeoutMs
• loadEngine
• a11yMaxIssues
• a11yEngine
• perfPreset
• perfEngine
• visualMode
• soakDurationSeconds
• soakIntervalSeconds
• autostartTarget

Legacy Fields (Not in run override)

Legacy fields browser/platform/device/headless/timeout/env are intentionally excluded.

Resources

• Generic skill scaffold: docs/skills/prooflane-mcp/SKILL.md
• uiq://runs/latest/manifest
• uiq://runs/latest/summary
• uiq://review/latest-release-brief
• uiq://manual-gates/inbox-summary

This site is open source. Improve this page.