aiwg/agentic/code/addons/uat-mcp/skills/uat-mode/SKILL.md

Deployment tool and support utility for AI context. Copies agents, skills, commands, rules, and behaviors into the paths each AI platform reads (Claude Code, Codex, Copilot, Cursor, Warp, OpenClaw, and 6 more) so one source of truth works across 10 platfo

---
namespace: aiwg
name: uat-mode
description: Detect requests for UAT generation, execution, or reporting and invoke the appropriate UAT command
version: 1.0.0
platforms: [all]

---

# UAT Mode

Detects requests related to User Acceptance Testing of MCP tool surfaces and routes to the appropriate UAT command.

## Triggers


Primary phrases matched automatically from skill description. No additional alternate expressions defined.

## Detection Logic

### High Confidence (Auto-Suggest)

Trigger contains explicit UAT terminology or MCP testing intent:
- Contains "UAT" in any case
- Contains "acceptance test" + "MCP" or "tool" or "server"
- Contains "test" + "MCP tools" or "MCP connections"

### Medium Confidence (Suggest with Alternatives)

Trigger is about testing tools but not explicitly UAT:
- "validate the tools"
- "check MCP coverage"
- "test the server"

### Low Confidence (Don't Suggest)

Too generic to confidently route:
- "run tests" (could be unit tests)
- "check coverage" (could be code coverage)
- "test this" (too vague)

## Behavior

### When No UAT Plan Exists

1. Inform user no plan exists
2. Offer to generate one: "I'll generate a UAT plan first. Run `/uat-generate`?"
3. If user agrees, invoke `/uat-generate`
4. After generation, offer to execute: "Plan ready. Execute it?"

### When UAT Plan Exists But No Results

1. Show existing plan summary (phases, test count)
2. Offer to execute: "Found existing plan. Execute it?"
3. If user agrees, invoke `/uat-execute`

### When Results Exist

1. Show latest results summary (pass/fail counts)
2. Offer options:
   - View full report: `/uat-report`
   - Re-run tests: `/uat-execute`
   - Generate new plan: `/uat-generate`

### Full Pipeline Request

When user says "acceptance test this server" or similar:
1. Generate plan: `/uat-generate`
2. Human reviews plan
3. Execute plan: `/uat-execute`
4. Generate report: `/uat-report`

## Response Templates

### Plan Generated
```
UAT plan generated with {N} tests across {M} phases.
Review the plan at: {path}

Ready to execute? Use `/uat-execute {path}` or just say "run the UAT".
```

### Execution Complete
```
UAT execution complete:
  Passed: {pass}/{total} ({percentage}%)
  Failed: {fail}
  Issues filed: {count}

View full report with `/uat-report` or say "show UAT results".
```

### No MCP Servers
```
No MCP servers detected. UAT-MCP requires at least one connected MCP server.
Check your MCP configuration and try again.
```

## Integration Notes

- **Priority**: High (when UAT keywords detected)
- **Exclusivity**: Full (takes over the interaction when triggered)
- **Fallback**: If UAT addon not installed, suggest `aiwg use uat-mcp`

## Related

- Commands: `/uat-generate`, `/uat-execute`, `/uat-report`
- Agents: `uat-planner`, `uat-executor`
- Issue: #380

## References

- @$AIWG_ROOT/agentic/code/addons/aiwg-utils/rules/vague-discretion.md — Measurable UAT completion criteria
- @$AIWG_ROOT/agentic/code/frameworks/sdlc-complete/README.md — SDLC framework context for acceptance testing
- @$AIWG_ROOT/docs/cli-reference.md — CLI reference for uat commands
- @$AIWG_ROOT/agentic/code/addons/aiwg-utils/rules/native-ux-tools.md — Interactive question patterns for UAT flow