← 返回 Skills 市场
ariffazil

Geox Ground

作者 ariffazil · GitHub ↗ · v1.0.0 · MIT-0
cross-platform ✓ 安全检测通过
86
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install geox-ground
功能描述
Provides physics-based, data-grounded Earth science reasoning with explicit uncertainty for geology, wells, seismic, petrophysics, and subsurface analyses.
使用说明 (SKILL.md)

GEOX Ground — Grounded Earth Reasoning Layer

GEOX exists to force the agent to reason through physics, material constraints, real data, and explicit uncertainty when the task touches the physical Earth.

The GEOX Habit

When any Earth-domain claim is made, the agent must:

  1. Separate epistemic levels — never conflate observation with interpretation
  2. Invoke physics first — appeal to material constraints, not narrative
  3. Preserve uncertainty — don't paper over unknowns with confident language
  4. Ground in real data — wells, seismic, logs, samples beat inference

Epistemic Labels

Prefix every Earth-domain claim with a label:

Label Meaning Example
OBS Direct observation "Wireline log shows 2150m TVD"
DER Derived from data "Porosity computed from density log"
INT Interpretation "This structure likely contains hydrocarbons"
SPEC Speculation "Possibly a turbidite fairway"

Never present INT as OBS. Never present SPEC as INT.

When This Skill Activates

  • Geology, geophysics, petrophysics, basin modeling
  • Well logs, seismic interpretation, formation evaluation
  • Rock properties, fluid contacts, pressure regimes
  • Structural geology, stratigraphy, sedimentology
  • Questions with words: "formation", "reservoir", "trap", "seal", "source", "pay", "porosity", "permeability", "saturation", "pressure", "depth", "structure"

GEOX Decision Rules

  1. Physics over narrative — "The seal capacity must exceed the column height" beats "it looks like a good trap"
  2. Real wells over hand-wavy synthesis — cite actual data points
  3. Uncertainty bands required — for any quantitative claim, provide a range
  4. Hold conditions — if data is insufficient for a confident interpretation, say so and label it SPEC
  5. No false precision — 2150m ± 5m is fine; 2150.342m is fraud if you only measured to nearest 5m

Anti-Error Patterns

❌ "This is definitely a turbidite reservoir." ✅ "The seismic character in this interval (OBS: chaotic, moderate amplitude) is consistent with turbidite deposits (INT), but could also be mass-transport deposits (DER from amplitude vs frequency analysis). Confidence: LOW."

❌ "The well found commercial hydrocarbons." ✅ "The MDT tool recovered gas-cut mud (OBS) indicating mobile gas (DER: pressure gradient 0.65 psi/m). Pay thickness net pay cutoffs pending final log analysis."

References

  • GEOX epistemic label guide: references/epistemic-labels.md
  • Common hold conditions in geoscience: references/hold-conditions.md
安全使用建议
This skill is internally consistent and appears safe as an instruction-only grounding layer: it doesn't install software, request credentials, or reach out to external endpoints. Two practical cautions: (1) source/homepage is missing — you cannot verify the author or maintainers, so prefer skills with a known provenance if that matters to you; (2) the skill asks the agent to use and cite real well/seismic data — avoid pasting proprietary or confidential datasets into the agent unless you are comfortable sharing them. Finally, treat its outputs as advisory: for safety-critical or operational decisions, validate with qualified domain experts and original data sources.
功能分析
Type: OpenClaw Skill Name: geox-ground Version: 1.0.0 The skill bundle contains instructions and reference guides for a 'GEOX Ground' reasoning layer designed to improve the accuracy and rigor of geoscience-related AI responses. It mandates the use of epistemic labels (OBS, DER, INT, SPEC) and defines safety-critical 'hold conditions' for geological data interpretation. There is no executable code, no network activity, and no evidence of malicious intent or prompt injection aimed at subverting the agent's security.
能力标签
cryptocan-make-purchases
能力评估
Purpose & Capability
Name/description (physics- and data-grounded geoscience reasoning) match the SKILL.md and reference docs. No unrelated env vars, binaries, or installs are requested.
Instruction Scope
SKILL.md prescribes labeling (OBS/DER/INT/SPEC), uncertainty bands, and hold conditions for geoscience claims. It does not instruct reading system files, accessing external endpoints, or exfiltrating data; it only biases agent reasoning and output format.
Install Mechanism
Instruction-only skill with no install spec and no code files — nothing is written to disk or fetched at install time.
Credentials
No environment variables, credentials, or config paths are requested. The guidance to cite 'real data' refers to domain data (wells, logs) and does not imply access to system secrets.
Persistence & Privilege
Skill is not always-on and does not request elevated privileges or modify other skills. Model invocation is allowed (default), which is expected for skills; this alone is not a risk here.
如何使用
  1. 确保已安装 OpenClaw(本地或 Docker 部署)
  2. 在对话框中输入安装命令:/install geox-ground
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /geox-ground 触发
  4. 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
版本历史
v1.0.0
Initial APEX-level release: Earth reasoning grounding with OBS/DER/INT/SPEC epistemic labels, physics-over-narrative discipline
元数据
Slug geox-ground
版本 1.0.0
许可证 MIT-0
累计安装 0
当前安装数 0
历史版本数 1
常见问题

Geox Ground 是什么?

Provides physics-based, data-grounded Earth science reasoning with explicit uncertainty for geology, wells, seismic, petrophysics, and subsurface analyses. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 86 次。

如何安装 Geox Ground?

在 OpenClaw 或 Claude Code 对话框中运行命令「/install geox-ground」即可一键安装,无需额外配置。

Geox Ground 是免费的吗?

是的,Geox Ground 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。

Geox Ground 支持哪些平台?

Geox Ground 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。

谁开发了 Geox Ground?

由 ariffazil(@ariffazil)开发并维护,当前版本 v1.0.0。

💬 留言讨论