replicant

Clone and search external repositories locally for source-grounded code research.

2|2|Updated Dec 23, 2017
One-click install
npx skills add https://github.com/xriu/dotfiles --skill replicant-xriu
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: replicant
Source: https://github.com/xriu/dotfiles/tree/main/home/.skills/replicant
Command: npx skills add https://github.com/xriu/dotfiles --skill replicant-xriu

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Answers about open-source libraries often rely on stale documentation or web snippets instead of actual source code. This Skill maintains durable, human-findable local clones of external repositories so answers are grounded in real implementation evidence with commit SHAs and file paths. ## Core Features & Use Cases - Local-first repo resolution: Searches an existing clone shelf and inventory file before any web search, normalizing URLs, owner/repo names, and package names to a consistent clone path layout. - Managed clone lifecycle: Clones missing repos with configurable transport (HTTPS/SSH) and depth (shallow/full), and updates existing clones according to a configurable policy that protects dirty working trees. - Evidence-based answers: Records commit SHA and working tree state before use, then cites file paths and line ranges from actual source code. - Use Case: When asked how a routing library handles navigation internally, the Skill locates or clones the repo, inspects the source and tests, and answers with the exact commit and file references. ## Quick Start Ask how a specific feature is implemented in a GitHub repository and have the answer grounded in its actual source code.

Frequently Asked Questions about replicant

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

FAQPage Schema
How do I research a GitHub repository's source code locally?▼

Clone the repository into a durable local directory organized by host/owner/repo, then search and read the source directly with tools like ripgrep. Record the commit SHA before use so answers can cite exact code versions and file paths.

How to find a repository when I only know the package name?▼

Check the package metadata first, for example with npm view <package> repository.url, then search GitHub with gh search repos if needed. Never invent a repo mapping; ask the user when the match is ambiguous.

Should I use a shallow or full git clone for code research?▼

A shallow clone with --depth 1 is fast and small, and works for reading current source. A full clone is better when you need history, blame, or tag-heavy work such as version-specific questions.

What happens if a local clone has uncommitted changes?▼

Dirty clones are not updated automatically. The working tree state is reported and the user is asked before any pull, and destructive commands like reset --hard or clean -fd require explicit approval.

Can I check the source code of a specific version or tag?▼

Yes. Fetch tags with git fetch --tags, check out the matching tag, and inspect the source at that revision. The user is told about the detached HEAD state and the previous branch is restored afterward when practical.