Skip to content

RULES.en.md

Latest commit

 

History

History
102 lines (77 loc) · 3.1 KB

RULES.en.md

File metadata and controls

102 lines (77 loc) · 3.1 KB
Raw

🌐 English | 中文

Bounty Plaza Rules

Core Principle

Quality first, speed as tiebreaker. Best quality code wins. At equal quality, first submitter wins.

Task Lifecycle

Find GitHub bounty
       |
       +---> External: AI generates initial fix -> submit PR (lock bounty)
       |
       +---> Internal: Post task Issue on bounty-plaza
                              |
                    Contributors submit fixes -> auto score (>=90)
                              |
                    First to pass wins
                              |
                    Force-push winning code to the PR
                              |
                    Upstream merges -> bounty paid -> coins issued
  1. Find → Admin finds a GitHub bounty Issue
  2. Lock → AI generates a working fix in 5 min → fork upstream → submit PR
  3. Distribute → Post task on bounty-plaza with description, reward, and rules
  4. Compete → Contributors submit fixes via PR → auto score ≥90 → first to pass wins
  5. Cover → Admin force-pushes winning code to the upstream PR
  6. Wait → Upstream reviews and merges
  7. Settle → Upstream merges + bounty received → coins issued to winner
  8. Redeem → Winners can cash out coins at any time

⚠️ Never submit an empty PR. First version must be real working code (AI-generated). Force-push updates later is normal.

Risk Coverage

Scenario Response
Someone else merged first Close PR, cancel task
Reviewer requests changes Include fixes when force-pushing winning code
No one scores ≥90 in 72h Submit AI version (no coins issued)

Submission Rules

  • All fixes via Pull Request to bounty-plaza
  • Each PR must include:
    • ✅ Working fix code
    • ✅ Explanation of the fix
    • ✅ Test cases
  • Admin triggers scoring

Scoring

Dimension Weight Max How
🧪 Correctness 40% 40 pytest: all pass = 40, any fail = 0
🔒 Security 35% 35 AST scan + bandit: -7 per violation
📐 Code Quality 15% 15 pylint score / 10 × 15
⚡ Performance 10% 10 Runtime vs baseline

Passing: ≥ 90 points

Veto Items (automatic 0)

  • Hardcoded expected output (AST detected)
  • Tampered test cases
  • Banned modules: pickle/marshal/ctypes/eval/exec
  • Dangerous syscalls: os.system, subprocess.Popen
  • Empty or gibberish code
  • HIGH/MEDIUM bandit findings

Pricing

Bounty Range R (Your Share) Platform Fee
< $50 95% ≈14%
$50 - $200 92% ≈17%
$200 - $500 90% ≈19%
$500 - $1000 87% ≈22%
$1000+ 85% ≈24%

Formula: Your Cash = Bounty × 1.25 × R × 0.72

See REWARD_POLICY.en.md for full details.

Prohibited

  • ❌ Plagiarism
  • ❌ Malicious code
  • ❌ Multiple accounts on same task
  • ❌ Irrelevant submissions
  • ❌ Tampering with test cases

💡 Quality first, zero tolerance on cheating. Fix well + fix fast + don't cheat = winner.