Extract HARA function mappings from 功能清单 tables into structured JSON.
npx skills add https://github.com/WSJ40/hara-mutiagent --skill hara-stage0
Please help me install this Agent Skill. Skill: hara-stage0 Source: https://github.com/WSJ40/hara-mutiagent/tree/main/skills/hara-stage0 Command: npx skills add https://github.com/WSJ40/hara-mutiagent --skill hara-stage0
hara-stage0 turns a HARA “功能文档” (or its source_extraction JSON) into a strictly structured, deterministic function mapping file, so downstream stages can perform analysis using only the extracted function scope rather than guessing from risk knowledge.
output/<RUN_ID>_stage0_function_mapping.json, optionally generating per-function context files for faster downstream processing.Run stage0 from a normalized function document by executing: python tools/hara/generate_stage0_function_mapping.py --input <function_doc_path> --out output/<RUN_ID>_stage0_function_mapping.json --run-id <RUN_ID> --write-contexts
High-intent search queries and answers about installing and using this skill.
To extract function lists into JSON, this Skill deterministically reads the “功能清单” table and binds each listed function name to matching正文标题 content for downstream pipeline analysis.
Deterministic function mapping in a HARA pipeline strictly extracts only the functions listed in the “功能” table column, ensuring downstream stages analyze exact functional scope rather than guessing from risk knowledge.
You generate per-function context files by executing the stage0 generation script with the --write-contexts flag, which bundles matched功能标题 and child content into detail_text raw aggregation.
Stage0 function extraction requires YAML frontmatter discovery fields, strict table/column rules, continuous Function_ID numbering, and validated UTF-8 JSON output to process functional documents correctly.
The best way to preprocess functional documents is using a controlled HARA pipeline stage0 script, which avoids inventing functions from正文标题 and preserves original technical wording in detail_text aggregation.
Function extraction output may miss names if the source document lacks strict table/column rules or continuous Function_ID numbering, as the script only extracts headers exactly matching “功能”.
cogni-workValidate SKILL.md files and extract metadata into standardized YAML documents.
WSJ40Convert Stage 1 safety faults into vehicle-level hazard JSON entries per Function_ID.
ianyimiEnforce coding standards for the global tech stack in Claude Code.
WSJ40Derive Stage 1 function-level malfunction candidates from Stage 0 mappings.