Introduction

In the DSH / Agent development scenario, when an agent “searches GitHub,” one typically doesn’t just want a list of repository names, but rather results that have been filtered, deduplicated, sorted by relevance, and output as summaries suitable for entering the context window.

Fectivnfy112357/github-explore is a DSH plugin for AI coding agents, positioned as “Discovery + management wrappers around the gh CLI for AI coding agents”. Below is an introduction to its features, installation methods, and typical usage.

What is this

Fectivnfy112357/github-explore is maintained by Fectivnfy112357 and is licensed under MIT.

It provides GitHub discovery and management wrappers around the gh CLI, wrapping gh search and gh repo view, without introducing new CLI entry points. It requires the gh CLI and Python 3.10+.

Core Features

The verified plugin capabilities are listed below:

  • Multi-axis exploration: supports parallel axes, relevance scoring, canonical anchor recall, awesome-list signals.
  • Smart defaults: filters forks and archived repos by default, enforces a minimum star threshold, deduplicates by fullName, and renders a layered markdown summary.
  • Layered output: summary output to stdout, full report output to a temporary file.
  • --schema support: applicable to find_repos, explore, repo_summary; other scripts mirror gh search’s camelCase fields.
  • No new CLI surface: capabilities are all wrapped around gh search and gh repo view.

Installation and Activation

If installed via the DSH plugin method, use the repository shorthand:

dsh plugin --profile web add Fectivnfy112357/github-explore

The plugin registers a skill via ctx.skills at runtime.

Before use, ensure the gh CLI is authenticated:

gh auth status

If you no longer need it, you can remove the plugin:

dsh plugin --profile web remove github-explore

Typical Usage

The following commands are from verified usage examples. Before execution, ensure the current working directory contains the corresponding scripts/ directory and that gh is authenticated.

Search repositories by language and minimum star threshold:

python scripts/find_repos.py "vector database" --language python --min-stars 500

Multi-axis Exploration

Define multiple search axes for the same topic and merge the results:

python scripts/explore.py "multi-agent" --axis "framework|multi-agent framework in:readme; collaborative agents in:readme" --axis "protocol|A2A agent protocol in:readme; agent-to-agent communication in:readme"

Machine-readable Output

When machine-readable output is needed, explicitly pass --format json; the pipe does not automatically switch output format.

Viewing Output Structure

--schema supports the following scripts:

  • find_repos
  • explore
  • repo_summary

Output Location

The plugin outputs the summary to stdout and automatically writes the full report to a temporary file:

%TEMP%/gh-explore-{topic}-{ts}.md

Applicable Scenarios and Notes

Suitable for scenarios where GitHub project discovery, summarization, and JSON output need to be integrated into the DSH agent workflow.

Notes before use:

  • Depends on the gh CLI; authentication must be completed first.
  • Requires Python 3.10+.
  • The plugin runs with the permissions of the current DSH process; it is recommended to review the source code and license before installing.
  • License is MIT.
  • Output is layered: stdout summary and temporary file full report.

Links

GitHub Repository:

https://github.com/Fectivnfy112357/github-explore