Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add GGbond-bo/MemOmics-Agent --skill debate-coregit clone --depth 1 https://github.com/GGbond-bo/MemOmics-AgentWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/ggbond-bo/memomics-agent/debate-core)<a href="https://agentmods.dev/skills/ggbond-bo/memomics-agent/debate-core"><img src="https://agentmods.dev/badge/skills/ggbond-bo/memomics-agent/debate-core/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/ggbond-bo/memomics-agent/debate-core"><img src="https://agentmods.dev/badge/skills/ggbond-bo/memomics-agent/debate-core.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00138 | $0.02113 |
| Opus 5 | $0.00069 | $0.01056 |
| Sonnet 5 | $0.00028 | $0.00423 |
| Haiku 4.5 | $0.00014 | $0.00211 |
Grade A, and why
debate-core scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 5d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 118 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Debate Core: 何时辩、怎么辩、怎么回流
辩论是 MemOmics 的质量内核:结论、参数、可入库知识都必须过辩论。 但不是每次都辩——全量辩论有害(iMAD, AAAI 2026 Oral:选择性触发省 92% token 且准确率反升 13.5%)。用门控决定何时辩。
1. 什么时候该辩论(debate_gate 三级门控)
enforcement.py 的 debate_gate(es, stage, signals) 自动判定,返回
(level, reasons, force):
| 级别 | 含义 | 成本 |
|---|---|---|
| L0 | 跳过(chat/lightweight 级,无分析对象) | 0 |
| L1 | 轻量(单对正反+裁判 / 3 采样投票) | ≈1/4 |
| L2 | 完整 8 角色(正方3+反方4+裁判) | 全量 |
五类触发信号(命中即升级):
- 高影响(high_impact):入库/报告/结论产物工具(generate_report、 save_knowledge、add_figure 等)→ 强制 L2,不可降级
- 失败(failed_retries≥2 或 last_error):同命令重试≥2 次 → L2
- 冲突(conflict):rail_review(post) 未通过 / 与上次结果差异大 → L2
- 不确定性(uncertainty):候选参数≥2、措辞犹豫、自评低置信 → 结论前 L2
- 阶段(stage):analysis 级结论合成前默认 L2;脚本设计/执行后默认 L1
级别默认值:chat/lightweight → L0(不辩);statistical → L1; analysis → 结论前 L2、其余 L1。
预算护栏:单会话辩论次数 ≥ config debate.budget(默认 3)后,
非强制 L2 自动降 L1。topic 级去重:同一主题只辩一次
(debated_topics 集合,替代旧的 debate_done 布尔)。
L2 裁判 confidence=low 且要入库 → 自动升级重辩(不可用低置信结论入库)。
2. 怎么辩论(引擎参数化,config.yaml debate: 段)
debate_analysis(topic, context, mode=?, rounds=?, role_model_map=?):
不传参数时全部从 config 读,config 无 debate 段 = 现状行为(兼容)。
| mode | 含义 | 适用 |
|---|---|---|
| homogeneous | 单模型 8 角色(默认/现状) | 日常 L1/L2 |
| adversarial | 正/反/判三组异构模型 | 实验、高争议 |
| multi_model | 每个角色独立模型 | 实验、多样性最大 |
| temperature | 同模型多温度采样 | 对照实验 |
rounds:轮数,>1 时第 2 轮起向正反方注入上一轮裁判摘要 (角色依然看不到彼此原始论点——隔离不破坏)role_model_map:角色级模型覆盖,优先级最高 (角色名:pro_biology/pro_statistics/pro_bioinformatics/ con_biology/con_statistics/con_bioinformatics/con_history/judge)- 缓存指纹:缓存 key = md5(topic+context+mode指纹)。不同架构 永不共享缓存结果。改 mode/rounds/role_model_map 必然是新辩论。
- 模型解析优先级:role_model_map[label] → mode 分组 (adversarial 的 judge/pro/con;multi_model 按角色哈希从 provider_keys 分配)→ 环境变量(_sync_debate_env 注入的 _current_model)
- 无环境 key 时回退 provider_keys.json:跳过失效 dcs-cloud,优先 deepseek 官方(deepseek-v4-flash),其余兜底
v2(2026-08-27)新增参数(config 的 debate 段或工具参数):
- prompt_version: 2(默认):v2 证据契约 + 结构化论据 + 裁判 rubrics; MEMOMICS_DEBATE_LEGACY_PROMPTS=1 一键回退 v1 提示词
- evidence_cards:外部证据卡(JSON 数组或文本,含 PMID/DOI/effect/n/source_file), 注入所有角色与裁判;内容变化 → 指纹变化 → 不复用缓存
- role_preset: core7 | core9:core9 增加实验设计评审与可重复性评审两个中立角色 → 10 角色/轮
- judge_count: 1 | 3:多裁判温度采样 + 简单多数投票(judge_consensus 字段);
1 时成本 ×2-3,预算护栏按调用次数计
- rounds_max: 5:rounds 上限护栏(默认 5)
- max_tokens: {judge:8192, role:2048, l1_role:2048, l1_judge:8192} —— 推理模型(deepseek-v4-pro)若上限过低会被 reasoning 吃满,content 为空 → 裁决恒低
- 并发说明(2026-08-14 修正):受控并发 max_workers=3(MEMOMICS_DEBATE_MAX_WORKERS), 不再全串行(会卡死)也不全并发(会触发配额 8/8 失败);隔离性不受影响
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 5d ago First seen · 118 lines · 138 tokens per session scan A 2195bb1c2f19
debate-core is a skill published in the GitHub repository GGbond-bo/MemOmics-Agent (19 stars, last pushed 6d ago), licensed MIT. It adds 138 tokens to every session and 2,113 once invoked, about $0.0007 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-03.
Other skills, from other repositories
code-review-standards
Severity-tagged code review rubric (CRITICAL/HIGH/MEDIUM/LOW) used by the code-critic agent to produce APPROVE/WARN/BLOCK verdicts with evidence-backed findings.
instrument-data-to-allotrope
Convert laboratory instrument output files (PDF, CSV, Excel, TXT) to Allotrope Simple Model (ASM) JSON format or flattened 2D CSV. Use this skill when scientists need to standardize instrument data for LIMS systems, data lakes, or downstream analysis. Supports auto-detection of instrument types. Outputs include full…
exploratory-data-analysis
Perform bounded, local exploratory analysis of explicitly supported scientific files. Use for redacted CSV/TSV/JSON profiles; optional NumPy, HDF5, FASTA/FASTQ, and basic image metadata inspection; missingness/leakage audits; outlier and transformation sensitivity; and rigorous EDA report scaffolds. Other domain…
matlab
Build, review, migrate, and safely plan MATLAB or GNU Octave numerical workflows, including arrays, tabular/time data, tests, projects, graphics, MAT files, and explicit Python interoperability.
phylogenetics
Build and analyze phylogenetic trees using MAFFT (multiple alignment), IQ-TREE 2 (maximum likelihood), and FastTree (fast NJ/ML). Visualize with ETE3 or FigTree. For evolutionary analysis, microbial genomics, viral phylodynamics, protein family analysis, and molecular clock studies.
research-engineer
An uncompromising Academic Research Engineer. Operates with absolute scientific rigor, objective criticism, and zero flair. Focuses on theoretical correctness, formal verification, and optimal implementation across any required technology.