Stats
Actions
Tags
From exodia
Runs tournament of N competing AI agents (default 5) on coding task $0; judge scores solutions by correctness/elegance/performance with penalties; implements winner.
How this command is triggered — by the user, by Claude, or both
Slash command
/exodia:tournamentThis command is limited to the following tools:
The summary Claude sees in its command listing — used to decide when to auto-load this command
# TOURNAMENT — Competitive Coding with Penalty Scoring > N competitors. 1 judge. Winner's code ships. **Task:** $0 **Competitors:** $1 (default: 5) --- ## SCORING (Visible to ALL) | Criterion | Points | Penalty | Points | |-----------|--------|---------|--------| | Correctness | 40 | Style nitpicks | -2 | | Elegance | 25 | Over-engineering | -3 | | Performance | 20 | Unnecessary complexity | -3 | | Completeness | 15 | Doesn't compile | -10 | | | | False claims | -5 | **Tiebreaker:** Correctness → Performance → First submitted --- ## TEAM ARCHITECTURE --- <CRITICAL_EXECUTION_REQU...
N competitors. 1 judge. Winner's code ships.
Task: $0 Competitors: $1 (default: 5)
| Criterion | Points | Penalty | Points |
|---|---|---|---|
| Correctness | 40 | Style nitpicks | -2 |
| Elegance | 25 | Over-engineering | -3 |
| Performance | 20 | Unnecessary complexity | -3 |
| Completeness | 15 | Doesn't compile | -10 |
| False claims | -5 |
Tiebreaker: Correctness → Performance → First submitted
TOURNAMENT JUDGE (You — Orchestrator)
│
├─ Round 1: COMPETITION ($1 parallel competitors)
│ ├── competitor-1 ... competitor-N
│ └── GATE → all complete
│
├─ Round 2: JUDGING (1 agent)
│ └── tournament-judge
│ └── GATE → winner selected
│
└─ Round 3: IMPLEMENTATION (1 agent)
└── winner-implementer
└── Build + Test → SHIPPED | BLOCKED
<CRITICAL_EXECUTION_REQUIREMENT>
STEP -1 — Inherit Prior Findings:
If <EXODIA_FINDINGS_CONTEXT> tag exists in session context, read .eight-gates/artifacts/findings.json.
Inject relevant findings into competitor prompts as shared starting context — level the playing field.
YOU ARE THE JUDGE, NOT A COMPETITOR.
YOUR NEXT MESSAGE: $1 Task tool calls. NOTHING ELSE.
</CRITICAL_EXECUTION_REQUIREMENT>
Launch $1 agents in ONE message. Identical prompt per competitor:
subagent: feature-dev:code-architect | model: opus
TOURNAMENT — You are competing against others on the SAME task. Only the BEST wins.
TASK: $0
SCORING: Correctness(40) + Elegance(25) + Performance(20) + Completeness(15) = 100 PENALTIES: Style(-2), Over-engineering(-3), Complexity(-3), No-compile(-10), False claims(-5) TIEBREAKER: Correctness → Performance → First submitted
Write the BEST solution. Code must compile and pass tests. AVOID penalties. Explain WHY your solution should win.
Output: All code changes + approach explanation + why this wins
subagent: feature-dev:code-reviewer | model: opus
JUDGE $1 solutions for: $0
SCORING: Correctness(40) + Elegance(25) + Performance(20) + Completeness(15) PENALTIES: Style(-2), Over-engineering(-3), Complexity(-3), No-compile(-10), False claims(-5) TIEBREAKER: Correctness → Performance → First submitted
For each: score criteria, apply penalties, calculate total.
Output: SCORECARD: | Competitor | Correct | Elegant | Perf | Complete | Penalties | TOTAL | PENALTY LOG: Competitor X: -N (reason) FINAL RANKING with winner's solution code
subagent: feature-dev:code-architect
IMPLEMENT the winning tournament solution. Apply all code changes. Ensure tests pass. Format properly. Output: Files changed + verification
Run build and test using the project's toolchain.
+====================================================================+
| TOURNAMENT RESULTS |
+====================================================================+
| Task: $0 | Competitors: $1 |
+--------------------------------------------------------------------+
| 1st: [name] — [score]/100 |
| 2nd: [name] — [score]/100 |
| 3rd: [name] — [score]/100 |
+--------------------------------------------------------------------+
| Winner Applied: YES | Build: PASS/FAIL | Tests: PASS/FAIL |
+====================================================================+
npx claudepluginhub ancplua/ancplua-claude-plugins --plugin exodia/startGenerates and executes agent team orchestration from any task prompt. Supports --dry-run, --auto-run, --team-size, --models, approval, and output flags.
/competitive-analysisCompares this plugin against external plugins, tools, feature sets, or ideas using URL, path, or description input. Produces structured gap analysis report with specs to close gaps.
/kasi-multiDispatches N specialist agents in parallel to tackle the mission and synthesizes their reports into a single user-facing response. Defaults to N=6; supports sudo for min-2 fast mode.
/test-teamPaper tests code files with a 3-agent team (Happy Path Validator, Edge Case Hunter, Red Team Attacker) that debates findings and produces a prioritized flaw report.