360 Secure SkillHub
360 安全技能中心

extropy

面向 Extropy CLI 的执行型 Skill：按 spec→scenario→persona→sample→network→simulate→results 流程运行仿真实验，各阶段设质量门禁，先校验上游产物再执行下游，失败时诊断并升级处理，最终依据路径、SQL 与指标证据输出仿真分析结果，并配套运维、排障、分析与报告模板文档。

deveshparagiri
deveshparagiri
热度 32.05
最近更新 2026-05-11 17:27:04
分类：数据依赖：代码执行 / 读写文件 / 第三方 API应用场景：数据处理 / 报告撰写 / 研究 / 工作流

安装技能

复制下面的话术发给你的 Agent，它会通过 secure-skills 检索并安装该技能。

执行命令安装此 skill： npx secure-skills-cli add extropy
Metadata
name
extropy
description
Execution-first operator for Extropy: run pipelines, diagnose failures, and deliver evidence-backed simulation analysis using current CLI contracts.
allowed-tools
Read, Grep, Glob, Bash, Edit, Write
argument-hint
[goal or experiment request]
always
false
source
https://github.com/exaforge/extropy
homepage
https://github.com/exaforge/extropy
required-binaries
["extropy"]
required-env-vars
["OPENAI_API_KEY","ANTHROPIC_API_KEY","OPENROUTER_API_KEY","DEEPSEEK_API_KEY","AZURE_API_KEY","AZURE_ENDPOINT","AZURE_OPENAI_API_KEY","AZURE_OPENAI_ENDPOINT","MODELS_FAST","MODELS_STRONG","SIMULATION_FAST","SIMULATION_STRONG","SIMULATION_MAX_CONCURRENT","SIMULATION_RATE_TIER","SIMULATION_RPM_OVERRIDE","SIMULATION_TPM_OVERRIDE"]
required-config-files
["~/.config/extropy/config.json"]
primary-credential
Provider API keys via environment variables
required_binaries
["extropy"]
required_env_vars
["OPENAI_API_KEY","ANTHROPIC_API_KEY","OPENROUTER_API_KEY","DEEPSEEK_API_KEY","AZURE_API_KEY","AZURE_ENDPOINT","AZURE_OPENAI_API_KEY","AZURE_OPENAI_ENDPOINT","MODELS_FAST","MODELS_STRONG","SIMULATION_FAST","SIMULATION_STRONG","SIMULATION_MAX_CONCURRENT","SIMULATION_RATE_TIER","SIMULATION_RPM_OVERRIDE","SIMULATION_TPM_OVERRIDE"]
required_config_files
["~/.config/extropy/config.json"]
primary_credential
Provider API keys via environment variables
metadata
{"requires":{"binaries":["extropy"],"config_files":["~/.config/extropy/config.json"],"environment_variables":["OPENAI_API_KEY","ANTHROPIC_API_KEY","OPENROUTER_API_KEY","DEEPSEEK_API_KEY","AZURE_API_KEY","AZURE_ENDPOINT","AZURE_OPENAI_API_KEY","AZURE_OPENAI_ENDPOINT","MODELS_FAST","MODELS_STRONG","SIMULATION_FAST","SIMULATION_STRONG","SIMULATION_MAX_CONCURRENT","SIMULATION_RATE_TIER","SIMULATION_RPM_OVERRIDE","SIMULATION_TPM_OVERRIDE"]},"credentials":{"primary":["provider API keys via environment variables only"],"notes":["extropy reads model/provider settings from ~/.config/extropy/config.json","do not read raw .env contents unless explicitly requested by the user"]}}

Extropy Operator

Run experiments end to end, with strict quality gates and reproducible commands.

Operating Rules

  1. Execute commands; do not only describe them.
  2. Validate upstream artifacts before expensive downstream stages.
  3. Use explicit evidence (paths, SQL, metrics) for every conclusion.
  4. Keep assumptions visible and minimal.
  5. Escalate after repeated failure of the same gate.

Runtime Dependencies and Credential Scope

  • Required binary: extropy must be installed and on PATH.
  • Config file read: ~/.config/extropy/config.json.
  • Credentials expected: provider API keys from env vars (OPENAI_API_KEY, ANTHROPIC_API_KEY, OPENROUTER_API_KEY, DEEPSEEK_API_KEY, AZURE_API_KEY/AZURE_OPENAI_API_KEY).
  • Azure endpoint vars: AZURE_ENDPOINT or AZURE_OPENAI_ENDPOINT when using Azure providers.
  • Safety boundary: only access credentials/config needed to execute extropy commands for the requested study/scenario, do not inspect raw .env values, and avoid reading unrelated files.

Canonical Pipeline

extropy spec -> extropy scenario -> extropy persona -> extropy sample -> extropy network -> extropy simulate -> extropy results

Quick Runbook

STUDY=runs/my-study
SCENARIO=ai-shock

# 1) Study + base population
extropy spec "5000 US working-age adults" -o "$STUDY" --use-defaults
cd "$STUDY"
extropy config set cli.mode agent

# 2) Scenario + persona
extropy scenario "AI systems outperform most knowledge workers in 6 months" -o "$SCENARIO" -y
extropy persona -s "$SCENARIO" -y

# 3) Sample + network
extropy sample -s "$SCENARIO" -n 5000 --seed 42 --strict-gates
extropy network -s "$SCENARIO" --seed 42 --quality-profile strict --validate

# 4) Simulate
extropy simulate -s "$SCENARIO" --seed 42 --fidelity high --rpm-override 1000

# 5) Results + raw extraction
extropy results -s "$SCENARIO" summary
extropy results -s "$SCENARIO" timeline
extropy query states --to states.jsonl

Required Quality Gates

  • spec: extropy validate population.vN.yaml passes.
  • scenario: extropy validate scenario/<name>/scenario.vN.yaml passes.
  • persona: persona file exists and validates.
  • sample: count correct, coherence gates pass, no critical impossibles.
  • network: no isolated catastrophes, topology metrics in acceptable range.
  • simulate: run completes or checkpoints cleanly with recoverable state.

Recommended Command Surface

  • Build: spec, scenario, persona, sample, network, simulate
  • Inspect: results summary|timeline|segment|agent
  • Deep checks: query summary|network|network-status|sql|agents|edges|states
  • Agent QA: chat list, chat ask

Module Map

  • OPERATIONS.md: stage-by-stage execution and rerun strategy.
  • TROUBLESHOOTING.md: failure diagnosis and escalation policy.
  • ANALYSIS.md: post-run metrics, SQL patterns, reporting structure.
  • SCENARIOS.md: scenario/pipeline design patterns and limits.
  • REPORT_TEMPLATE.md: standardized delivery format.