sddeval-collect

Collect code changes from an SDD branch into evaluation JSON.

52|3|Updated Feb 14, 2026
One-click install
npx skills add https://github.com/ZTE-AICloud/Co-OmniSpec --skill sddeval-collect
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: sddeval-collect
Source: https://github.com/ZTE-AICloud/Co-OmniSpec/tree/main/skills/sddeval-collect
Command: npx skills add https://github.com/ZTE-AICloud/Co-OmniSpec --skill sddeval-collect

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

SDDEval collects code change information after an SDD workflow, extracting feature_infos and code_blocks and generating the JSON file required for evaluation, saved to the current SDD branch's evalset directory.

Core Features & Use Cases

  • Automatically identifies the current branch and the corresponding tasks.md to drive data extraction.
  • Extracts feature_infos (the "目的" lines) from tasks.md and builds code_blocks from diffs in the target directory.
  • Writes a structured evaluation JSON to changes/<branch>/evalset/configiEval.generated.json for downstream evaluation and testing.

Quick Start

Run the skill on the current SDD branch to collect code changes and generate the evalset JSON.

Frequently Asked Questions about sddeval-collect

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

FAQPage Schema
How do I automatically collect code changes for SDD evaluation?

You can collect SDD evaluation data by running the skill on your current branch, which reads tasks.md, extracts code blocks from git diffs, and saves a generated JSON file to the evalset directory.

What is the JSON schema structure for generated SDD evaluation data?

Generated SDD evaluation JSON files include api and input sections, where feature_infos are extracted from tasks.md and code_blocks are derived from git diffs in the target directory.

Do I need a git repository to generate SDD evaluation datasets?

A git repository is required because the skill identifies the current SDD branch, reads tasks.md, and analyzes code diffs in the target directory to generate the evaluation JSON.

How to extract feature_infos from tasks.md for code change evaluation?

Feature_infos are extracted by reading the '目的' lines from tasks.md in the changes/<branch>/ directory, which are then used to build the structured evaluation JSON payload.

Where does the generated evaluation JSON file get saved in an SDD workflow?

The evaluation JSON file is saved to changes/<branch>/evalset/configiEval.generated.json, containing api and input sections structured for downstream evaluation and testing.

Can I specify a custom target directory for code diff analysis?

You can specify a custom target directory for diff analysis, but the skill defaults to 'networking_zte' if none is provided, analyzing diffs to build code_blocks for the evaluation JSON.