Analysis updated 2026-05-18
Ask your AI assistant to research an external library by cloning it locally and reading the actual source code.
Get answers about open-source project internals backed by a specific commit, file, and line range.
Turn an approved architecture or domain document into typed domain models and service boundaries before writing feature code.
| oscabriel/skills | 709166872-cpu/tagcast-ai | advdebug/brovan | |
|---|---|---|---|
| Stars | 51 | 51 | 51 |
| Language | — | HTML | C# |
| Setup difficulty | easy | moderate | moderate |
| Complexity | 2/5 | 4/5 | 5/5 |
| Audience | developer | data | developer |
Figures from each repo's GitHub metadata at analysis time.
Requires an AI coding assistant that supports skills, installed via the skills.sh CLI.
This repository contains a small collection of agent skills, which are instruction files that AI coding assistants can follow as slash commands. The repository is designed to be installed as a unit via a tool called skills.sh, which makes all included skills available to your AI assistant in one step. The first skill is called Replicant. It instructs an AI agent to research external code repositories by cloning them locally rather than relying on web search, cached documentation, or generated summaries. When you ask about a library, framework, or open-source project, Replicant tells the agent to check a local folder of clones first, clone the repository if it is missing, and then read the actual source files to answer your question. Configuration options include where to store clones, whether to update existing ones automatically, and whether to use full or shallow history. Answers from Replicant include the specific commit, file path, and line range the information came from. The second skill is called Docs to Types. It is meant for the moment after a project has produced design decisions, domain definitions, or architecture records, but before any feature code is written. The skill instructs an agent to turn that approved context into typed code structure: domain types, error types, service boundaries, state models, smart constructors, and module layout. It is not intended to produce business logic or full implementations. The readme recommends running a separate grilling session first to produce the input context documents. Both skills are aimed at developers who work closely with AI coding assistants and want structured, repeatable behavior for specific research and architecture tasks. The repository itself contains no application code.
A small pack of AI coding assistant skills: one that researches libraries by cloning and reading their real source, and one that turns approved design docs into typed code scaffolding.
The README does not state a license.
Setup difficulty is rated easy, with roughly 5min to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Verify against the repo before relying on details.