task-completition-evaluation

name	task-completition-evaluation
description	Final completion gate for VibeTeam tasks. Use at the end of implementation to verify diff quality, real testing, GitHub/Slack multi-agent communication evidence, and PR health before declaring done.

Task Completition Evaluation

Use this skill only at the end of a task, after implementation is complete and before reporting completion.

Required Inputs

• Task reference: GitHub issue/PR URL.
• Code evidence: git diff/patch and changed file list.
• Test evidence: command outputs and report files.
• Collaboration evidence: Slack thread URL(s), GitHub issue/PR/discussion URL(s), and eval report paths.

Completion Checklist (All Required)

1. Review diff/patch.
   • Run git status --short and git diff --stat.
   • Inspect full diff for correctness, scope control, and accidental edits.
2. Reflect: cleanup.
   • Remove dead code, debug output, TODO placeholders, and temporary files.
   • Confirm no tests were weakened to force passing.
3. Reflect: optimality.
   • Verify the issue is solved in the simplest robust way.
   • Confirm docs and design alignment for behavior changes.
4. Run real tests (not mocked shortcuts).
   • Export environment first:
     • export $( < ~/.env.d/codex.env )
     • export $( < .env )
   • Run unit/integration tests relevant to the touched area.
   • If agent behavior/routing/tools/eval logic changed, run at least one Slack eval.
   • Fail the checklist if tests were skipped without explicit justification.
5. Run GitHub evaluation tests for agent collaboration.
   • Use scripts/eval_github_e2e.py scenarios relevant to the task.
   • Verify and report:
     • different GitHub App identities participated as different agents
     • correct existing @githubapphandle mentions were used
     • agents communicated/handoff occurred across thread(s)
     • if assignment is non-assignable for role bots, report Assignment fallback mode evidence and bot replies after trigger
6. Run Slack evaluation tests for agent collaboration.
   • Use scripts/eval_slack_e2e.py scenarios relevant to the task.
   • Verify and report:
     • different Slack app identities participated as different agents
     • correct existing @slackapphandle mentions were used
     • agents communicated/handoff occurred
     • Slack thread URL is included and manually reviewed against requirements
     • transcript evidence is copied from the generated eval report Conversation History section (real run only)
   • For github_issue_pr_handoff_slack, require post-check pass for:
     • Slack required roles responded
     • Slack distinct role app identities
7. Create PR and verify checks.
   • Create a PR linked to the issue (Fixes #<issue>).
   • Confirm required GitHub checks pass before completion claim.

Minimum Command Set

# Inspect changes
git status --short
git diff --stat
git diff

# Test/eval environment
export $( < ~/.env.d/codex.env )
export $( < .env )

# Example unit tests
uv run python -m pytest tests/ -v

# Example Slack eval
uv run python scripts/eval_slack_e2e.py --scenario github_issue_pr_handoff_slack --channel C0ALG01DLJV --timeout 600

# Example GitHub eval
uv run python scripts/eval_github_e2e.py --scenario github_issue_pr_handoff_github --repo VibeTechnologies/vibeteam-eval-hello-world --pr 1 --actor-login OpenCodeEngineer --issue-role software_engineer --issue-assignee 'vibeteam-swe-bot-260301[bot]' --timeout 600

Evidence Required In Final Report

• Docs referenced:
  • docs/testing.md
  • docs/design.md
  • docs/requirements.md
• Design decisions applied: decision, reason, impact.
• Test results: exact commands + pass/fail summary.
• Eval artifacts:
  • Slack thread URL(s) and transcript summary.
  • GitHub thread URL(s).
  • Report file paths in results/eval_reports/.
  • At least one quoted message per required role from the report transcript (no placeholders).
• Clear verdict: COMPLETED only when every checklist item is satisfied.