mcpField
C
code-search

code-search

github.com/atholaDevelopment34385 installsUpdated 1d ago
Repository

About code-search

code-search is a skill for finding and ranking GitHub implementations and libraries related to a topic. It builds search queries, parses repository metadata into findings, and can optionally use the GitHub API for richer details.

Key features

  1. 01Builds GitHub search queries
  2. 02Runs searches through WebSearch
  3. 03Parses repository metadata
  4. 04Optionally retrieves GitHub API details
  5. 05Ranks findings by relevance

Use cases

  1. 01Finding existing implementations of a topic
  2. 02Discovering relevant GitHub libraries
  3. 03Supporting a tome research session

Install

Copy the command and run it in a terminal.

$npx skills add https://github.com/athola/claude-night-market --skill code-search

After installing, type/code-searchin Claude Code to confirm it loaded.

Details

From the project README

Code Search

When To Use

  • Finding existing implementations or libraries on GitHub
  • Part of a /tome:research session or standalone search

When NOT To Use

  • Searching local codebase (use Grep or Explore agent)
  • Academic literature (use Skill(tome:papers))

Search GitHub for implementations of a given topic.

Usage

Invoked as part of /tome:research or standalone.

Workflow

  1. Build search queries using tome.channels.github.build_github_search_queries()
  2. Execute queries via WebSearch
  3. Parse results via parse_github_result() (returns None for a result with no URL, which the caller drops)
  4. Optionally use GitHub API via build_github_api_search() for richer metadata
  5. Rank via rank_github_findings()
  6. Return Finding objects

Exit Criteria

  • At least one GitHub search query built and executed via WebSearch for the requested topic
  • Results parsed into Finding objects containing repo metadata (name, URL, stars) and a relevance score
  • Findings returned from rank_github_findings() sorted by relevance score descending
  • If no results are found for the query, this is reported explicitly rather than returning an empty list silently

Comments

Markdown supported