rlm-explore

Explores large codebases and dense context by extracting evidence-cited working sets.

1|Updated Apr 3, 2026
One-click install
npx skills add https://github.com/TierOne-Studio/spa-velocity --skill rlm-explore-tierone-studio
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: rlm-explore
Source: https://github.com/TierOne-Studio/spa-velocity/tree/main/.ruler/skills/rlm-explore
Command: npx skills add https://github.com/TierOne-Studio/spa-velocity --skill rlm-explore-tierone-studio

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? When facing a large or unfamiliar codebase, or when a user pastes dense material like logs, multiple documents, or long error output, loading everything into context at once is wasteful and error-prone. This Skill applies a disciplined five-step exploration method that inspects material in slices and produces a compact, evidence-backed Working Set to drive decisions. ## Core Features & Use Cases - Five-Operation Workflow: LOCATE relevant slices, EXTRACT minimal snippets, CHUNK large context, TRANSFORM into a Working Set of 5-15 bullets, and VERIFY against requirements. - Evidence-Cited Working Set: Every bullet in the output must cite a file:line, log line, or doc anchor, preventing unsupported claims. - REPL Transcript Discipline: Produces exact copy-pasteable ripgrep commands with expected findings, preferring symbol search over broad text search. - Use Case: A developer inherits an unfamiliar repository and needs to find where a config value is read. The Skill runs scoped searches, extracts only the relevant functions, and returns a verified Working Set instead of dumping whole files. ## Quick Start Explore this repository to find where the authentication token is validated and summarize the findings as an evidence-cited working set.

Frequently Asked Questions about rlm-explore

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

FAQPage Schema
How do I explore a large unfamiliar codebase efficiently?

Use a five-step method: locate relevant slices by keywords and symbols, extract minimal snippets, chunk large context, transform findings into a 5-15 bullet working set, and verify against requirements. Prefer scoped ripgrep searches over reading whole files.

How to search code with ripgrep instead of grep?

Use rg with scoped paths and symbol-oriented patterns rather than broad text searches across the whole repo. Avoid recursive searches through node_modules or build artifacts, and prefer symbol search over plain text matching when possible.

When should I not use recursive codebase exploration?

Skip it for small focused changes where the relevant code is already obvious, factual single-file questions, and trivial bug fixes. The method is designed for large or dense context, not quick lookups.

What is a working set in code exploration?

A working set is 5-15 bullets summarizing verified findings, where every bullet cites an evidence pointer such as a file:line, log line, or document anchor. No bullet is allowed without a source, which keeps conclusions grounded.

When should code exploration stop?

Stop when the working set fully covers the requirement, further searching duplicates earlier findings, a blocking ambiguity surfaces that needs user input, or confidence reaches roughly 0.9 that the right slices were found.