skills.rest
arndvs
arndvsCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

explore

Investigates codebases by spawning parallel sub-agents for deep multi-file exploration.

1|Updated Mar 26, 2026
Search Tags:#investigation#code audit#codebase exploration#sub-agents#architecture mapping#parallel search
One-click install
npx skills add https://github.com/arndvs/ctrlshft --skill explore-arndvs
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: explore
Source: https://github.com/arndvs/ctrlshft/tree/main/skills/explore
Command: npx skills add https://github.com/arndvs/ctrlshft --skill explore-arndvs

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Single-pass codebase exploration typically reads only 6-10 files, producing incomplete understanding of features, bugs, or architecture. This Skill decomposes investigation topics and runs multiple parallel sub-agents to achieve far deeper file coverage. ## Core Features & Use Cases - Parallel Sub-Agent Exploration: Decomposes a topic into narrow focus areas and spawns a dedicated explore sub-agent for each, running them in parallel. - Deep Traversal Mode: Uses the search_subagent tool with the explore keyword to trigger recursive directory and file searching, covering 20-30+ tool calls per agent. - Unified Synthesis: Merges all sub-agent findings into a single cohesive summary with cross-references between areas. - Use Case: When asked "how does authentication work?", it spawns separate agents to explore session creation, middleware route protection, token issuance, and logout handling, then synthesizes the results. ## Quick Start Ask the agent to explore how a specific feature works in this codebase, such as "explore how the authentication flow works".

Frequently Asked Questions about explore

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

FAQPage Schema
How do I explore a large codebase with an AI agent?▼

Decompose the topic into narrow areas of concern and spawn a dedicated explore sub-agent for each one in parallel. Each agent performs deep recursive file traversal, and the results are synthesized into a single unified summary.

How to investigate how a feature works in an unfamiliar repository?▼

Trigger the explore workflow with a question like "how does authentication work?". The skill splits it into focused sub-investigations such as session storage, middleware, and token handling, then cross-references the findings.

Why use multiple sub-agents instead of one agent for code exploration?▼

A single agent pass typically reads only 6-10 files, giving an incomplete picture. Dedicated explore sub-agents each call 20-30+ tools and aggressively search the repo, maximizing coverage and depth.

When should I not use parallel codebase exploration?▼

Avoid it for trivial lookups where the answer lives in one or two known files, since spawning multiple sub-agents adds overhead. It is designed for non-trivial investigations, audits, and architecture mapping.

What happens if a sub-agent finds something unexpected during exploration?▼

The workflow directs the agent to spawn an additional follow-up sub-agent to investigate the unexpected finding. All results are then merged into the final unified summary with cross-references.

Related Skills

Explore More Software Engineering Skills→
leonanpereirapintoleonanpereirapinto
1

investigar-area

Investigate a codebase area using a project map and scoped task.

Community
Intermediate
WangK0221WangK0221

byterover

Manages project knowledge via a persistent ByteRover context tree with a command-line interface.

Community
Advanced
testdoubletestdouble
198

refactor

Refactor code iteratively with test-gated validation after each step.

Official
Advanced
thethirdoriginthethirdorigin

codegraph

Index and query local codebases as semantic knowledge graphs.

Official
Advanced