hara-stage4

Derive Stage 4 SG_Sum summaries from validated HARA results with tool-based grouping.

2|Updated May 13, 2026
One-click install
npx skills add https://github.com/WSJ40/hara-mutiagent --skill hara-stage4
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: hara-stage4
Source: https://github.com/WSJ40/hara-mutiagent/tree/main/skills/hara-stage4
Command: npx skills add https://github.com/WSJ40/hara-mutiagent --skill hara-stage4

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Stage 4 SG_Sum 汇总把已通过 Stage 3 校验的 HARA 结果整理成可交付的 stage4 输出,同时严格保证仅由模型填写“操作模式”,其余字段全部由工具派生并可校验回溯。

Core Features & Use Cases

  • Stage 4 只补“操作模式”:从 generate_stage4_sg.py 生成的草稿中读取 SG_Sum 行与证据,仅让模型填写“操作模式”字段。
  • 严格派生与门禁校验:不允许模型改动安全目标、ASIL Level、安全状态、FTTI(ms)与Comments,且输出不包含 MF_ID。
  • 支持汇总规则落地:同一 MF 内相同安全目标汇总时,ASIL 取最高、FTTI(ms)取最小;来源 MF 通过 Comments 追溯。

Quick Start

Run generate_stage4_sg to create the Stage 4 SG_Sum draft, fill only the 操作模式 field based on Comments evidence, then validate the result with check_stage_json for stage4.

Frequently Asked Questions about hara-stage4

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I generate Stage 4 SG_Sum summaries from validated HARA results?

To generate Stage 4 SG_Sum summaries, run generate_stage4_sg to create a draft, fill only the operation mode field based on Comments evidence, then validate the output with check_stage_json for stage4 compliance.

What fields require deterministic tool derivation in Stage 4 SG_Sum?

Deterministic tool derivation is required for SG_No, safety goals, ASIL Level, safe state, FTTI(ms), and Comments. The model is restricted to filling only the operation mode field to ensure review-grade checking.

How does Stage 4 SG_Sum aggregate safety goals within the same MF?

When aggregating safety goals within the same MF, the process takes the highest ASIL level and the minimum FTTI(ms) value. The source MF is traced through the Comments field for review-grade checking.

Why does Stage 4 SG_Sum validation fail when MF_ID is present in the output?

Stage 4 SG_Sum validation fails when MF_ID is present because the output format strictly excludes MF_ID. Validation also enforces that operation_mode contains a non-placeholder value.

Can I manually edit ASIL Level or FTTI values during Stage 4 SG_Sum generation?

No, manual edits to ASIL Level or FTTI values are not allowed. These fields are strictly derived by the tool from Stage 3 validated HARA results to maintain deterministic traceability.

What is the JSON output file naming convention for Stage 4 SG_Sum results?

The JSON output file naming convention for Stage 4 SG_Sum results is output/<RUN_ID>_stage4_sg_sum.json, ensuring review-grade checking for each specific run.