---

description: Run user acceptance testing on a scoped set -- reads REVIEW-SCOPE.md allowed-tools: Read, Write, Bash, Agent, AskUserQuestion, Glob, Grep args:

• name: set type: set-ref description: Set to run user acceptance testing on required: true categories: [autonomous]

---

/rapid:uat -- User Acceptance Testing

You are the RAPID UAT skill. This skill runs user acceptance testing on a scoped set. It reads REVIEW-SCOPE.md (produced by /rapid:review) as its input. UAT runs ONCE on the full scope -- it is never chunked or concern-scoped. Follow these steps IN ORDER. Do not skip steps. Do NOT include stage selection prompting, unit test logic, or bug hunt logic.

Step 0: Environment + Set Resolution

0a: Load environment

if [ -z "${RAPID_TOOLS:-}" ] && [ -n "${CLAUDE_SKILL_DIR:-}" ] && [ -f "${CLAUDE_SKILL_DIR}/../../.env" ]; then export $(grep -v '^#' "${CLAUDE_SKILL_DIR}/../../.env" | xargs); fi
if [ -z "${RAPID_TOOLS}" ]; then echo "[RAPID ERROR] RAPID_TOOLS is not set. Run /rapid:install or ./setup.sh to configure RAPID."; exit 1; fi

Display Stage Banner

if [ -z "${RAPID_TOOLS:-}" ] && [ -n "${CLAUDE_SKILL_DIR:-}" ] && [ -f "${CLAUDE_SKILL_DIR}/../../.env" ]; then export $(grep -v '^#' "${CLAUDE_SKILL_DIR}/../../.env" | xargs); fi
if [ -z "${RAPID_TOOLS}" ]; then echo "[RAPID ERROR] RAPID_TOOLS is not set. Run /rapid:install or ./setup.sh to configure RAPID."; exit 1; fi
node "${RAPID_TOOLS}" display banner uat

0b: Parse arguments

The user invokes this skill with: /rapid:uat <set-id> or numeric shorthand like /rapid:uat 1.

Resolve Set Reference

If <set-id> was provided, resolve it through the numeric ID resolver:

# (env preamble here)
RESOLVE_RESULT=$(node "${RAPID_TOOLS}" resolve set "<set-input>" 2>&1)
RESOLVE_EXIT=$?
if [ $RESOLVE_EXIT -ne 0 ]; then
  echo "$RESOLVE_RESULT"
  # Display the error message from the JSON and STOP
fi
SET_NAME=$(echo "$RESOLVE_RESULT" | node -e "d=JSON.parse(require('fs').readFileSync(0,'utf-8')); console.log(d.resolvedId)")

Use SET_NAME for all subsequent operations.

If <set-id> was not provided, use AskUserQuestion to ask:

• question: "Which set to run UAT for?"
• options: List available sets from STATE.json by running:

  node "${RAPID_TOOLS}" state get --all
  

Detect --post-merge flag

Check if the user invoked with --post-merge flag: /rapid:uat <set-id> --post-merge

If --post-merge is present, set POST_MERGE=true. Post-merge mode reads REVIEW-SCOPE.md from the post-merge artifact directory.

0d: Check Review State

If POST_MERGE=true: Skip this step. Post-merge reviews do not track pipeline state.

Check if UAT has already been completed for this set:

REVIEW_STATE=$(node "${RAPID_TOOLS}" review state "${SET_NAME}" 2>&1)

Parse the JSON output. If the response contains a stages array, find the uat entry.

If uat is already complete (status: "complete"), use AskUserQuestion to prompt:

• question: "UAT stage has already been completed for this set (verdict: {verdict}). What would you like to do?"
• options: ["Re-run UAT", "Skip and exit"]

If user chooses "Skip and exit": print "UAT stage already complete." and exit. If user chooses "Re-run UAT": continue to Step 1.

If scope is not complete (scope entry missing or status not "complete") and POST_MERGE is not set:

• Print error: "Cannot run UAT: scope stage has not been completed. Run /rapid:review {set-id} first."
• Exit. Do NOT continue.

If unit-test is not complete (unit-test entry missing or status not "complete") and POST_MERGE is not set:

• Print error: "Cannot run UAT: unit-test stage has not been completed. Run /rapid:unit-test {set-id} first."
• Exit. Do NOT continue.

If no review state exists and POST_MERGE is not set:

• Print error: "Cannot run UAT: no review state found. Run /rapid:review {set-id} first."
• Exit.

Step 1: Load REVIEW-SCOPE.md

Determine the scope file path:

1. If POST_MERGE=true (explicit --post-merge flag was provided): Use .planning/post-merge/{setId}/REVIEW-SCOPE.md directly.

2. If POST_MERGE is not set (no flag): Auto-detect by checking paths in order:

   • First, try standard path: .planning/sets/{setId}/REVIEW-SCOPE.md
   • If not found, try post-merge path: .planning/post-merge/{setId}/REVIEW-SCOPE.md
   • If found at the post-merge path, set POST_MERGE=true so downstream artifact writes (REVIEW-UAT.md, issue logging) use the post-merge directory.

Guard check: If neither path contains the file, display error and STOP:

[RAPID ERROR] REVIEW-SCOPE.md not found for set '{setId}'.
Checked: .planning/sets/{setId}/REVIEW-SCOPE.md
         .planning/post-merge/{setId}/REVIEW-SCOPE.md
Run `/rapid:review {setId}` first to generate the review scope.

Read the file content. Parse the <!-- SCOPE-META {...} --> JSON block to extract metadata:

• setId, date, postMerge, worktreePath, totalFiles, useConcernScoping

Step 2: Parse Scope Data

Parse the following sections from REVIEW-SCOPE.md:

Changed Files

Extract ALL file paths from the ## Changed Files table. UAT uses the full file list.

Dependent Files

Extract ALL file paths from the ## Dependent Files table.

Acceptance Criteria

Extract criteria from the ## Acceptance Criteria numbered list. These are the primary input for UAT.

Important: UAT operates on the FULL scope (all changed files + all dependent files). It does NOT split by concern groups or directory chunks. This is by design -- acceptance testing evaluates the set holistically.

Step 3: Load Context

Acceptance Criteria

The acceptance criteria extracted in Step 2 are the primary test basis. Each criterion originates from a wave's PLAN.md and is prefixed with [wave-N].

Optional Context

If the set has a CONTEXT.md file (.planning/sets/{setId}/CONTEXT.md), read it to understand the set's implementation decisions and domain context. This helps the UAT agent generate more targeted test scenarios.

Step 4: Spawn UAT Agent -- Test Plan Generation

Spawn a single rapid-uat agent with the full scope:

User acceptance testing for set '{setId}' -- Test Plan Generation.

## All Changed Files
{complete list of changed files}

## All Dependent Files
{complete list of dependent files}

## Acceptance Criteria
{numbered list of acceptance criteria with wave attribution}

## Set Context
{CONTEXT.md content if available, or 'No additional context available.'}

## Working Directory
{worktreePath from SCOPE-META, or cwd if post-merge}

## Instructions
Generate a comprehensive UAT test plan with detailed step-by-step human verification instructions for each acceptance criterion.

For each acceptance criterion:
1. Create one or more test scenarios
2. For each scenario, specify:
   - name: descriptive test name
   - criterion: which acceptance criterion it validates (e.g., "[wave-1] Criterion text")
   - steps: array of objects, each with:
     - instruction: specific human-actionable instruction (e.g., "Navigate to http://localhost:3000/login in your web portal")
     - expected: specific observable outcome (e.g., "The page displays a login form with email and password fields")
   - files: which source files are relevant

Group scenarios under the acceptance criterion they validate. Duplicate cross-cutting scenarios under each criterion they touch.

Return via:
<!-- RAPID:RETURN {"status":"COMPLETE","data":{"testPlan":[{"name":"...","criterion":"[wave-N] ...","steps":[{"instruction":"...","expected":"..."}],"files":["..."]}]}} -->

Step 5: Present Test Plan for Approval

Display the test plan summary:

--- UAT Test Plan ---
Set: {setId}
Total Scenarios: {count}
Total Steps: {totalSteps}

Scenario 1: {name}
  Criterion: {criterion}
  Steps: {step count} steps
  Files: {file list}

Scenario 2: {name}
  Criterion: {criterion}
  Steps: {step count} steps
  Files: {file list}

---------------------

Use AskUserQuestion to present the plan:

• question: "Approve the UAT test plan?"

• options: ["Approve", "Modify", "Skip"]

• Approve: Proceed to Step 6

• Modify: Allow user to request changes, re-spawn agent with modifications, re-display

• Skip: Skip to Step 7 (artifact writing) with empty results

Step 6: Human Verification Loop

This is the core interaction loop. The skill drives a sequential walk through ALL steps of ALL scenarios, collecting human verdicts one at a time.

Implementation

1. Flatten all scenarios' steps into a sequential list with metadata:

   {scenarioName, criterion, stepIndex, totalSteps, instruction, expected, files}
   

2. Initialize counters: passed = 0, failed = 0, skipped = 0, failures = []

3. For each step in the flattened list:

   a. Use AskUserQuestion:

   • question: "Step {globalIndex}/{totalGlobalSteps} -- {criterion}\n\n**Scenario:** {scenarioName}\n\n**{instruction}**\n\nExpected: {expected}\n\nDoes this pass?"
   • options: ["Pass", "Fail", "Skip", "Pass all remaining"]

   b. On Pass: increment passed, continue to next step.

   c. On Fail: increment failed, then prompt for severity:

   • Use AskUserQuestion:
     • question: "Step {globalIndex} FAILED.\n\nWhat severity level?"
     • options: ["Critical", "High", "Medium", "Low"]
   • Record the severity from the user's choice.
   • Build the failure object:

     {
       "id": "<setId>-uat-<globalIndex>",
       "criterion": "<criterion text>",
       "step": "<instruction>",
       "description": "Expected: <expected>. Step: <instruction>",
       "severity": "<userChoice lowercase>",
       "relevantFiles": ["<files array>"],
       "userNotes": "",
       "expectedBehavior": "<expected>",
       "actualBehavior": "Failed (human reported)"
     }
     

   • Push the failure object to failures array.
   • Continue to next step.

   d. On Skip: increment skipped, continue to next step.

   e. On Pass all remaining: set all remaining steps to passed (add remaining count to passed), break the loop. This is the escape hatch for large test plans.

Notes on AskUserQuestion

• AskUserQuestion only supports predefined options, not freeform text input.
• The userNotes field is set to empty string because freeform input is not available.
• The description field is auto-populated from the step's instruction and expected fields.
• The actualBehavior is set to "Failed (human reported)" as the default.

Step 7: Write REVIEW-UAT.md

Write the UAT results to:

• Standard mode: .planning/sets/{setId}/REVIEW-UAT.md
• Post-merge mode: .planning/post-merge/{setId}/REVIEW-UAT.md

This write is idempotent -- if REVIEW-UAT.md already exists, overwrite it.

Format:

# REVIEW-UAT: {setId}

## Summary
| Metric | Value |
|--------|-------|
| Total Scenarios | {count} |
| Passed | {passed} |
| Failed | {failed} |
| Skipped | {skipped} |
| Human Verified | {passed + failed + skipped} |

## Criteria Coverage
| Criterion | Scenarios | Status |
|-----------|-----------|--------|
| [wave-1] Criterion text | 2 | PASS |
| [wave-2] Criterion text | 1 | FAIL |

## Scenario Results

### PASS: {scenario name}
- **Criterion:** {criterion}
- **Steps:** {passed}/{total}

### FAIL: {scenario name}
- **Criterion:** {criterion}
- **Steps:** {passed}/{total}
- **Failed Steps:**
  - Step 3: {step description} -- {error}

## Failed Scenarios Detail

### {scenario name}
- **Criterion:** {criterion}
- **Error:** {error description}
- **Relevant Files:** {file list}

Step 7b: Write UAT-FAILURES.md

Only if failures.length > 0. If there are no failures, do NOT write UAT-FAILURES.md (no file = no failures).

Write to:

• Standard mode: .planning/sets/{setId}/UAT-FAILURES.md
• Post-merge mode: .planning/post-merge/{setId}/UAT-FAILURES.md

If re-running and a previous UAT-FAILURES.md exists, overwrite it (clean overwrite).

The content follows the format locked in Wave 1:

# UAT-FAILURES

<!-- UAT-FORMAT:v2 -->

<!-- UAT-FAILURES-META {JSON object with "failures" array} -->

## Failures

### {id}: {criterion}
- **Step:** {step instruction}
- **Severity:** {severity}
- **Description:** {description}
- **User Notes:** {userNotes, if non-empty}

The <!-- UAT-FAILURES-META --> block contains a JSON object with a failures array. Each failure object has these fields:

• id: Unique identifier (<setId>-uat-<stepIndex>)
• criterion: The acceptance criterion text
• step: The step instruction that failed
• description: Combined description ("Expected: {expected}. Step: {instruction}")
• severity: One of critical, high, medium, low
• relevantFiles: Array of relevant source file paths
• userNotes: Empty string (freeform input not available via AskUserQuestion)
• expectedBehavior: The expected field from the step
• actualBehavior: "Failed (human reported)"

Step 8: Log Failed Steps

For each failed UAT scenario, log an issue.

CLI Flags (recommended):

node "${RAPID_TOOLS}" review log-issue "{setId}" \
  --type "uat" \
  --severity "{severity from human's choice in Step 6}" \
  --file "{primary relevant file}" \
  --description "UAT failure: {scenario name} -- {failure detail}" \
  --source "uat"

Stdin JSON alternative:

echo '{"id":"<uuid>","type":"uat","severity":"{severity}","file":"{primary relevant file}","description":"UAT failure: {scenario name} -- {failure detail}","source":"uat","createdAt":"<iso-timestamp>"}' | \
  node "${RAPID_TOOLS}" review log-issue "{setId}"

The CLI flag interface auto-generates id and createdAt. The stdin JSON interface requires all fields including id and createdAt.

Use the severity the human selected in Step 6 for each failure. Do not apply a heuristic -- the human's verdict is authoritative.

If in post-merge mode, issues are logged to .planning/post-merge/{setId}/REVIEW-ISSUES.json.

Step 8b: Record UAT Completion

If POST_MERGE=true: Skip this step. Post-merge reviews do not track pipeline state.

Determine the verdict based on UAT results:

• All scenarios passed and all criteria covered: verdict is pass
• Some scenarios failed or criteria not fully covered: verdict is partial
• Critical failures or major criteria gaps: verdict is fail

Mark the UAT stage as complete:

node "${RAPID_TOOLS}" review mark-stage "${SET_NAME}" uat {verdict}

Step 9: Completion Banner

Print the completion banner:

--- RAPID UAT Complete ---
Set: {setId}{postMerge ? ' (post-merge)' : ''}
Scenarios: {passed}/{total} passed
Failed: {failed}
Skipped: {skipped}
Human Verified: {passed + failed + skipped}
Criteria Coverage: {coveredCriteria}/{totalCriteria}
Issues Logged: {issueCount}
Failures Logged: {failures.length}

Output: {path to REVIEW-UAT.md}

Next steps:
  /rapid:bug-hunt {setIndex}   -- Run adversarial bug hunt
  /rapid:review summary {setIndex} -- Generate review summary
----------------------------

Display the completion footer:

if [ -z "${RAPID_TOOLS:-}" ] && [ -n "${CLAUDE_SKILL_DIR:-}" ] && [ -f "${CLAUDE_SKILL_DIR}/../../.env" ]; then export $(grep -v '^#' "${CLAUDE_SKILL_DIR}/../../.env" | xargs); fi
if [ -z "${RAPID_TOOLS}" ]; then echo "[RAPID ERROR] RAPID_TOOLS is not set. Run /rapid:install or ./setup.sh to configure RAPID."; exit 1; fi
node "${RAPID_TOOLS}" display footer "/rapid:review summary {setIndex}" --breadcrumb "review [done] > unit-test [done] > bug-hunt [done] > uat [done]"

Then exit. Do NOT prompt for stage selection.

Important Notes

• UAT runs ONCE on full scope. Unlike unit tests and bug hunt, UAT is never chunked or concern-scoped. It evaluates the entire set holistically against acceptance criteria. This is by design -- acceptance criteria span the whole set.
• Acceptance criteria are the primary input. The test plan is generated from the acceptance criteria extracted from wave PLAN.md files. Each criterion maps to one or more test scenarios.
• All testing is human-driven. Each step is presented individually via AskUserQuestion. The "Pass all remaining" option provides an escape hatch for large test plans.
• Idempotent overwrite. Re-running /rapid:uat overwrites REVIEW-UAT.md and UAT-FAILURES.md. Previous results are not accumulated. Git history preserves previous versions.
• REVIEW-SCOPE.md is the sole input. This skill does not scope files itself -- it reads the scope produced by /rapid:review. If the scope is stale, re-run /rapid:review first.
• No stage selection. This skill runs UAT only. It does not prompt the user to select unit test or bug hunt stages.