← 返回 Skills 市场
anonymouscodemaker

ELPA

作者 AnonymousCodeMaker · GitHub ↗ · v1.0.0 · MIT-0
cross-platform ⚠ suspicious
257
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install elpa
功能描述
Orchestrate real ELPA-style ensemble forecasting workflows by triggering external sub-model training jobs (for example PyTorch/Prophet/TiDE/transformers), th...
使用说明 (SKILL.md)

ELPA

Overview

This skill does not train toy adapters. It triggers real sub-model training commands from your own training codebases and then builds ELPA routing/weights from real validation errors.

Default model pool is intentionally larger than 4 and can be expanded freely.

Workflow

  1. Prepare a training config JSON (see assets/elpa_train_template.json).
  2. Dry-run the command plan to verify all sub-model commands.
  3. Execute real sub-model training when resources are available.
  4. Prepare validation error inputs per model.
  5. Build ELPA ensemble policy JSON from those errors.

1) Prepare Config

Create a config based on assets/elpa_train_template.json.

  • Put your real training entrypoints in each model train_cmd.
  • Keep each model tagged as online or offline.
  • Add as many models as needed; ELPA is not limited to 4.

2) Dry-Run Plan (No Training)

python3 scripts/elpa_orchestrator.py \
  --config assets/elpa_train_template.json \
  --run-dir .runtime/elpa_run \
  --manifest-out .runtime/elpa_run/train_manifest.json

This prints and records the commands that would run, without training.

3) Execute Real Training

python3 scripts/elpa_orchestrator.py \
  --config /path/to/your_train_config.json \
  --run-dir .runtime/elpa_run \
  --manifest-out .runtime/elpa_run/train_manifest.json \
  --execute

Use this only in an environment that has the required ML dependencies and hardware.

4) Build ELPA Integration Policy

After each sub-model produces validation errors, run:

python3 scripts/elpa_integrator.py \
  --config /path/to/your_integrate_config.json \
  --output .runtime/elpa_run/elpa_policy.json

The output includes:

  • scores for each model from validation errors
  • online_weights and offline_weights
  • best_online_model and best_offline_model
  • ELPA control fields (beta, dirty_interval, amplitude_window, mutant_epsilon)

Model Scaling

To support more models, append model blocks in your config with:

  • unique name
  • group as online or offline
  • real train_cmd

No script changes are needed for adding models.

Files

  • scripts/elpa_orchestrator.py: real sub-model training command planner/executor
  • scripts/elpa_integrator.py: ELPA score/weight builder from validation errors
  • assets/elpa_train_template.json: >4-model real training template
  • assets/elpa_integrate_template.json: ELPA integration template
  • references/config-schema.md: config field reference and placeholders
安全使用建议
This skill is coherent for orchestrating real training, but it will execute whatever shell commands appear in the JSON config. Only use configs and train_cmd templates from trusted sources. Recommended precautions: - Always run a dry-run first and inspect the generated manifest and train_cmd strings before using --execute. - Use a dedicated, sandboxed environment (container or isolated VM) with controlled dataset and filesystem access for execution. - Set run-dir to a directory you control and review stdout/stderr logs for secrets or unexpected output. - Inspect any per-model "env" entries in configs to ensure they don't inject sensitive credentials into subprocesses. - Avoid running this skill with configs obtained from untrusted third parties; validate templates and placeholders to prevent accidental execution of destructive commands (e.g., via shell metacharacters).
功能分析
Type: OpenClaw Skill Name: elpa Version: 1.0.0 The skill bundle contains a script, scripts/elpa_orchestrator.py, which executes arbitrary shell commands constructed from a JSON configuration file using subprocess.run(shell=True). While this is consistent with the stated purpose of orchestrating ML training workflows, the use of unsanitized string formatting to build shell commands represents a significant command injection vulnerability. No evidence of intentional malice, data exfiltration, or unauthorized network activity was found.
能力评估
✓ Purpose & Capability
Name/description match the code and assets: the orchestrator builds a plan and (optionally) runs real sub-model training commands and the integrator computes weights from validation errors. No unrelated credentials, binaries, or external services are requested.
ℹ Instruction Scope
SKILL.md instructs the agent to dry-run and optionally execute training commands from a user-supplied JSON config; the code reads those configs, renders command templates, writes manifests and logs, and reads CSVs for errors. This stays within the described purpose but implies execution of arbitrary shell commands from the config — expected for this tool, but a significant operational risk if configs are untrusted.
✓ Install Mechanism
Instruction-only with included Python scripts; there is no install spec, no external downloads, and no package installation steps. Files are written under the chosen run-dir; nothing is pulled from external hosts.
✓ Credentials
The skill declares no required environment variables or credentials. Per-model configs may include an "env" object which the orchestrator will merge into the subprocess environment — this is appropriate for training jobs but means the config can inject environment values into executed processes.
✓ Persistence & Privilege
The skill does not request always:true and does not attempt to modify other skills. It writes manifests, logs, and model directories under the run-dir (default .runtime/elpa_runs/<timestamp> or user-specified path), which is expected behavior for an orchestrator.
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 elpa
  3. 安装完成后，直接呼叫该 Skill 的名称或使用 /elpa 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 release of the elpa skill, enabling orchestration of real ELPA-style ensemble forecasting workflows. - Supports triggering external sub-model training jobs using your own training scripts for frameworks like PyTorch, Prophet, and transformers. - Automatically computes ELPA ensemble weights and policies from real sub-model validation errors. - Provides scalable model pool support—no hard limit on the number of models. - Includes templates, orchestration scripts, and integration tools for production-ready ensemble pipelines.
元数据
Slug elpa
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

ELPA 是什么？

Orchestrate real ELPA-style ensemble forecasting workflows by triggering external sub-model training jobs (for example PyTorch/Prophet/TiDE/transformers), th... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 257 次。

如何安装 ELPA？

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

ELPA 是免费的吗？

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

ELPA 支持哪些平台？

ELPA 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 ELPA？

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

推荐 Skills
self-improving agent
Captures learnings, errors, and corrections to enable continuous improvement. Use when: (1) A command or operation fails unexpectedly, (2) User corrects Clau...
⬇ 463283 · by pskoett
Skill Vetter
Security-first skill vetting for AI agents. Use before installing any skill from ClawdHub, GitHub, or other sources. Checks for red flags, permission scope, and suspicious patterns.
⬇ 259410 · by spclaudehome
Polymarket
Query Polymarket prediction markets. Check odds, find trending markets, search events, track price movements.
⬇ 232503 · by joelchance
Self-Improving + Proactive Agent
Self-reflection + Self-criticism + Self-learning + Self-organizing memory. Agent evaluates its own work, catches mistakes, and improves permanently. Use when...
⬇ 200423 · by ivangdavila
Github
Interact with GitHub using the `gh` CLI. Use `gh issue`, `gh pr`, `gh run`, and `gh api` for issues, PRs, CI runs, and advanced queries.
⬇ 191113 · by steipete
ontology
Typed knowledge graph for structured agent memory and composable skills. Use when creating/querying entities (Person, Project, Task, Event, Document), linkin...
⬇ 190156 · by oswalpalash

💬 留言讨论