ralph-teams-codex 1.5.1

package/skills/teams-verify/SKILL.md

@@ -0,0 +1,162 @@
+---
+name: teams-verify
+description: "Guide the user through manually verifying the build E2E — walks through each scenario step-by-step, records pass/fail, and writes a verification report."
+user-invocable: true
+---
+
+# Teams: Manual Verify
+
+You guide the user through manually verifying the completed build end-to-end. One scenario at a time. You record their results and write a final report.
+
+---
+
+## Step 1: Load the Plan
+
+Read `.ralph-teams/PLAN.md`. If not found:
+> `.ralph-teams/PLAN.md` not found. Run `teams-plan` first.
+
+Extract:
+- Plan ID (the `Plan ID:` field — e.g. `#2`)
+- All tasks
+- Acceptance criteria
+- Verification scenarios
+
+Also read `.ralph-teams/REVIEW.md` if it exists (to know what was already flagged by the automated reviewer).
+
+---
+
+## Step 2: Setup Check
+
+Before starting, ask the user:
+
+> **Before we verify, confirm setup:**
+>
+> - Is the app running? (URL or device/simulator)
+> - Any test accounts or data needed?
+> - Anything you want to skip?
+
+Wait for their response. Note any skips.
+
+---
+
+## Step 3: Walk Through Scenarios
+
+For each verification scenario from `.ralph-teams/PLAN.md`, present it one at a time:
+
+```
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+  RALPH-TEAMS  Plan #[N] — Scenario [N of M]
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+  [Scenario name]
+
+  Steps:
+  1. [Step]
+  2. [Step]
+  3. [Step]
+
+  Expected result:
+  [What the user should see]
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+```
+
+Then ask:
+> **Result? `pass` / `fail` / `skip` — or describe what you saw.**
+
+- **pass** → record it, move to next scenario
+- **fail** → ask: *"What went wrong?"* Record their description. Then immediately offer:
+  > **Bug detected. Run `teams-debug` to fix it now, or continue verifying the rest first?**
+
+  If they say **fix now**: invoke the `teams-debug` skill directly, passing the scenario name and their failure description as context. After the fix is applied, re-run this scenario before continuing.
+
+  If they say **continue**: record the failure and move to the next scenario.
+- **skip** → record as skipped with reason, move on
+
+Keep going until all scenarios are covered.
+
+---
+
+## Step 4: Write Verification Report
+
+Write results to `.ralph-teams/VERIFY.md`:
+
+```markdown
+# Manual Verification Report: [Feature Name]
+
+Plan ID: #[N]
+Date: [date]
+Verified by: User
+
+## Summary
+- Total scenarios: N
+- Passed: N
+- Failed: N
+- Skipped: N
+
+## Results
+
+### ✓ Scenario 1: [Name]
+Status: PASS
+
+### ✗ Scenario 2: [Name]
+Status: FAIL
+User reported: [description]
+
+### — Scenario 3: [Name]
+Status: SKIPPED
+Reason: [reason]
+
+## Acceptance Criteria
+- [x] Criterion 1
+- [ ] Criterion 2 — FAILED (see scenario 2)
+```
+
+---
+
+## Step 5: Summary
+
+Print the final result:
+
+```
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+  RALPH-TEAMS  Plan #[N] — Done — [N passed, N failed, N skipped]
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+  ✓  Scenario 1: [name]
+  ✗  Scenario 2: [name]
+  —  Scenario 3: [name]  (skipped)
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+```
+
+If there are failures, offer:
+
+> **N scenario(s) failed. Options:**
+> - `teams-debug` — fix bugs one at a time with full plan context (recommended)
+> - **Fix all** — I spawn a single builder to address all failures at once
+
+If they choose **`teams-debug`**: invoke the `teams-debug` skill for each failed scenario in order, passing the scenario name and failure description. After each fix, mark the scenario as `FIXED` in `.ralph-teams/VERIFY.md`.
+
+If they choose **fix all**: read the failures from `.ralph-teams/VERIFY.md` and use `spawn_agent` to start a builder subagent:
+
+```
+spawn_agent(
+  agent_type: "worker",
+  model: "gpt-5.4-mini",
+  message: "You are fixing multiple bugs found during manual verification.
+
+    Failed scenarios:
+    [paste full list of failures from .ralph-teams/VERIFY.md]
+
+    Feature plan (.ralph-teams/PLAN.md):
+    [paste full PLAN.md content]
+
+    Instructions:
+    - Fix each failed scenario
+    - Platform: [web|mobile from PLAN.md]
+    - Verify fixes using [Playwright|Maestro] after applying them
+    - If verification tools are not available, run tests/lint instead
+    - Commit with message: 'fix: address verification failures'"
+)
+```
+
+Wait for the builder with `wait_agent`. After you have handled the result and updated any verification state, call `close_agent`.
+
+Then offer to re-run verification on just the failed scenarios.