# Verification Before Completion

> Run verification commands and confirm output before claiming work is complete. Evidence before assertions. Thin ytstack wrapper around the vendored superpowers verification-before-completion procedure, injecting the current task's verification command. Use before commit, before PR, before summarize-task, before any 'done' claim.

- Skill: `yesterday-ai/verification-before-completion` (Agent Skill)
- Install (CLI): `npx skillmds@latest add yesterday-ai/verification-before-completion`
- Raw SKILL.md: https://api.skillmd.com/api/skills/yesterday-ai/verification-before-completion/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Research & Search
- Author: Yesterday-AI (https://skillmd.com/u/yesterday-ai)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/yesterday-ai/verification-before-completion

---


# ytstack:verification-before-completion

Thin wrapper around the vendored superpowers `verification-before-completion` skill (`vendor/superpowers/skills/verification-before-completion/SKILL.md`). Injects the current task's verification command extracted from the task-plan, then inlines the vendored procedure verbatim. Vendor is single source of truth.

## ytstack context (resolved at render time)

```!
cd "${CLAUDE_PROJECT_DIR:-$PWD}" 2>/dev/null || true
_PROJECT_SLUG=$(basename "$(git rev-parse --show-toplevel 2>/dev/null)" 2>/dev/null || basename "$(pwd)")
if [ -d "$PWD/.ytstack" ]; then _YT_DIR="$PWD/.ytstack";
elif [ -d "$HOME/.ytstack/projects/$_PROJECT_SLUG" ]; then _YT_DIR="$HOME/.ytstack/projects/$_PROJECT_SLUG";
else _YT_DIR=""; fi
_M=none; _S=none; _T=none
if [ -n "$_YT_DIR" ] && [ -f "$_YT_DIR/STATE.md" ]; then
  _M=$(sed -n 's/^current_milestone: *//p' "$_YT_DIR/STATE.md" | head -1)
  _S=$(sed -n 's/^active_slice: *//p' "$_YT_DIR/STATE.md" | head -1)
  _T=$(sed -n 's/^active_task: *//p' "$_YT_DIR/STATE.md" | head -1)
fi
_TASK_PLAN="$_YT_DIR/$_M-$_S-$_T-PLAN.md"
_FENCE=$(printf '\140\140\140')
_VERIFICATION=""
if [ -f "$_TASK_PLAN" ]; then
  _VERIFICATION=$(awk -v f="$_FENCE" '
    /^## Verification/{in_section=1; next}
    /^## /{in_section=0}
    in_section && index($0,f)==1 { in_fence=!in_fence; next }
    in_section && in_fence { print }
  ' "$_TASK_PLAN" | head -5)
fi
echo "HAS_YTSTACK: $([ -n "$_YT_DIR" ] && echo yes || echo no)"
echo "ACTIVE_TASK: $_M-$_S-$_T"
echo "TASK_PLAN: $_TASK_PLAN"
echo "TASK_PLAN_EXISTS: $([ -f "$_TASK_PLAN" ] && echo yes || echo no)"
echo "VERIFICATION_CMD_FOUND: $([ -n "$_VERIFICATION" ] && echo yes || echo no)"
if [ -n "$_VERIFICATION" ]; then echo "VERIFICATION_CMD:"; echo "$_VERIFICATION"; fi
```

## ytstack invocation notes

The command(s) printed above are the current task's verification commands, extracted from the `## Verification` section of the task-plan. These are what the vendored procedure must actually execute and confirm output of; do not substitute a different "looks like it should work" command.

HARD-GATE: if `TASK_PLAN_EXISTS=no` or `VERIFICATION_CMD_FOUND=no`, abort with: "No verification command in the active task-plan. Either run `ytstack:plan-task` to add one, or invoke this skill outside a ytstack task context where the user will provide the command." Do not invoke the vendored procedure without a command to verify.

After the vendored procedure completes:
- If verification passed: the user may now run `ytstack:summarize-task`.
- If verification failed: do NOT auto-retry. Report raw output and hand control back. The user decides next step (fix-and-retry vs re-scope the task).

## Vendored procedure (inlined verbatim)

```!
cat "${CLAUDE_SKILL_DIR}/../../vendor/superpowers/skills/verification-before-completion/SKILL.md"
```