메뉴
Spawn an isolated math-checker subagent to independently verify that src/ code matches docs/MATHEMATICAL_FORMULAS.md. Use after any edit to fusion.py, model.py, interaction.py, or loss.py.
기본적으로 소스를 먼저 확인하는 Prompt가 선택됩니다. 직접 명령으로 전환하거나 로컬 사본을 다운로드할 수도 있습니다.
설치 여부를 결정하기 전에 SKILL.md와 SkillsMP에 표시된 보조 파일을 읽어 보세요.
SOC 직업 분류 기준
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
직접 명령은 검토 Prompt를 거치지 않습니다. 실행하기 전에 소스를 확인하세요.
npx skills add https://github.com/Cuki2910/M2G-NET --skill math-checker명령은 한 줄로 유지됩니다. 복사하기 전에 가로로 스크롤해 전체 내용을 확인하세요.
로컬 사본을 원하시나요? SkillsMP에서 현재 제공할 수 있는 파일을 다운로드하세요.
SKILL.md 표시 중
Standardized M2G-NET experiment pipeline: data check → train → evaluate → save results to docs/. Use when running training, ablation, baseline comparison, gate stability, or any multi-phase experiment on M2G-NET.
Validate that M2G-NET code (src/fusion.py, src/model.py, src/interaction.py) matches the documented math in docs/MATHEMATICAL_FORMULAS.md. Use whenever editing model architecture, loss, gate computation, calibration, or metrics.
Spawn an isolated result-analyst subagent to read docs/ and outputs/ and produce a structured experiment summary with metric tables, key findings, and next-step recommendations.
| name | math-checker |
| description | Spawn an isolated math-checker subagent to independently verify that src/ code matches docs/MATHEMATICAL_FORMULAS.md. Use after any edit to fusion.py, model.py, interaction.py, or loss.py. |
| tags | ["m2g-net","math","verification","subagent"] |
| version | 1.0.0 |
| user-invocable | true |
| allowed-tools | ["Agent"] |
Spawns an isolated subagent that independently verifies M2G-NET math implementation. The subagent has no access to the current conversation context — it gives a clean second opinion.
src/fusion.py, src/model.py, src/interaction.py, src/loss.pymath-validator skill flags a WARN and you want independent confirmationUse the Agent tool with the subagent definition:
Agent(
description="Independent math verification for M2G-NET",
subagent_type="general-purpose",
prompt=[contents of .agents/subagents/math-checker/agent.yaml default_prompt]
)
docs/MATHEMATICAL_FORMULAS.md vs. src/fusion.py — sparsemax, gate shapes, temperaturesrc/loss.py — mask m_k applied correctly (zero, not average)src/model.py — view concatenation order matches formula notationsrc/interaction.py — cross-level interaction dimensionspython scripts/verify_math.py — automated shape checksReturns a table: Formula | File:Line | ✅PASS / ⚠️WARN / ❌FAIL | Detail
A clean run should show all PASS. Any FAIL must be fixed before the next commit to src/.