evaluate

/evaluate — Deep Issue Evaluation

You are running the WondrAIWork evaluation phase (Phase 02). Your job is to deeply assess whether a specific issue is worth contributing to.

Step 1: Parse Arguments

The user should provide <owner/repo> <issue-number>. If missing, ask for them.

Example: /evaluate vercel/next.js 42846

Step 2: Run Evaluation Script

Execute the evaluation script:

bash ${CLAUDE_PLUGIN_ROOT}/scripts/evaluate.sh $ARGUMENTS

This produces a structured report with:

• Repository health metrics (stars, forks, language, license, activity)
• Issue details (title, state, labels, assignee, reactions, comments)
• Risk assessment (competing PRs, CLA requirements, archived status)
• Community PR merge speed (average days to merge for external contributors)
• Verdict (VIABLE / CAUTION / SKIP)

Step 3: Read Evaluation Methodology

Read the full evaluation framework:

cat ${CLAUDE_PLUGIN_ROOT}/methodology/02-evaluate.md

Step 4: Apply Go/No-Go Framework

Score each criterion as Pass/Fail:

#	Criterion	Check
1	Repo merged external PRs in last 3 months	From merge speed report
2	Issue acknowledged by maintainer	Check issue comments
3	No competing active PR	From competing PRs section
4	CI is green on default branch	gh run list --repo REPO --branch main --limit 3
5	Scope is estimable (hours, not weeks)	Your judgment based on issue
6	Required domain knowledge is available	Honest assessment
7	No disqualifying red flags	CLA issues, archived, etc.

Decision rules:

• All 7 pass → Strong GO
• 1-2 fail → Conditional GO (note risks)
• 3+ fail → NO-GO (return to discovery)

Step 5: Calculate Weighted Score

Apply the scoring matrix from Phase 01:

• Impact (40%): Based on reactions, comments, issue severity
• Feasibility (35%): Based on scope, competition, technical complexity
• Visibility (25%): Based on repo stars, org reputation

Note: This is the Phase 02 preliminary score. It MUST be re-evaluated after Phase 03 (Understand) when the actual codebase complexity is known.

Step 6: Check for AI Contribution Policy

Check if the target repo has a policy on AI-assisted contributions:

gh api "repos/$1/contents/CONTRIBUTING.md" --jq '.content' 2>/dev/null | base64 -d | grep -i "ai\|copilot\|llm\|gpt\|claude" || echo "No AI policy found"

If they restrict AI contributions, flag this immediately.

Step 7: Present Evaluation Summary

Present a structured summary:

=== EVALUATION SUMMARY ===

Repo:     owner/repo (★ stars)
Issue:    #number — title
Score:    X.X/5.0 (Impact: X | Feasibility: X | Visibility: X)
Verdict:  GO / CONDITIONAL GO / NO-GO

Go/No-Go Checklist:
[x] External PRs merged recently
[x] Issue acknowledged by maintainer
[ ] No competing PRs  ← RISK: 1 open PR found
...

Key Risks:
- ...

Recommendation:
- ...

Step 8: Suggest Next Steps

Based on the verdict:

If GO:

This issue looks viable. The next step is the Owner Briefing (Phase 02b). The owner briefing presents the evaluation to the human owner for approval before any implementation begins. Use /briefing to generate the briefing.

If CONDITIONAL GO:

This issue has risks: [list risks]. The owner should decide whether to proceed. Use /briefing to present the evaluation with these risks highlighted.

If NO-GO:

This issue is not recommended. Reasons: [list]. Use /discover to find alternative targets.

Important Notes

• Competition check timing: Always re-check for competing PRs right before Phase 04 (Implementation), not just here
• Feasibility re-scoring: The Phase 02 score is preliminary. After Phase 03 (Understand), re-score feasibility based on actual codebase complexity. If it drops below 4/10, abandon the target.
• Owner Briefing is mandatory: Never proceed to Phase 03 without owner approval (Phase 02b)