etherled

Rls — AI skill for Claude Code

AI community

Local code search for AI agents: one exe, CLI for humans + MCP for AI.

How to install Rls

This entry records only its repository, not the path inside it, so there is no exact command to give. Open etherled/rls and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Rls does

Local code search for AI agents: one exe, CLI for humans + MCP for AI.

Alternatives in AI

Peekaboo — Peekaboo is a macOS CLI & optional MCP server that enables AI agents to capture screenshots of applications, o 3.2k ★
OpenContext — A personal context store for AI agents and assistants—reuse your existing coding agent CLI (Codex/Claude/OpenC 1.1k ★
Codeseek — Rust-powered code intelligence CLI for AI coding agents 764 ★

README

RLS · Let AI read 90% less code, every hit verifiable

**Binary-only distribution (no source)** · 中文版：[README.zh-CN.md](README.zh-CN.md) · Free trial · Windows x64 first

RLS (Rust Local Search) runs locally: **one `rls.exe`, two interfaces** — humans use the CLI, AI agents (Claude Code / Codex, …) use HTTP MCP against the same index and result semantics.

# 60-second trial
.\rls.exe search "retry_with_backoff" --scope D:\your-project --compact
.\rls.exe outline D:\your-project\src --max-files 50
.\rls.exe context D:\your-project\src\retry.py 45

⭐ If it saves you tokens / file reads, please Star — for a closed-source project this is the public feedback channel. 🐟 Enterprise / on-prem / custom languages: see [CONTACT.md](CONTACT.md).

Why RLS?

AI coding cost is rarely "can't find one line" — it is **Grep → Read × N, refilling context and guessing call graphs**.

RLS collapses it to:

Locate symbol/file → inspect structure/definition → trace calls / impl / data-flow on demand

Return **file + symbol + line** first; let AI deep-read only what matters.

Five tools

Tool	Does	Note
`search`	exact literal locate of defs/refs	`def/ref/all`, no regex
`outline`	file/directory structure map	map first
`context`	expand enclosing definition at `file+line`	source + callers/callees
`trace`	call path / implementation / data-flow / compare	evidence graph, `--compact` paging
`handover`	project handover fact prefetch	one command to hand over

Details: [docs/FEATURES.md](docs/FEATURES.md), [docs/BEST-PRACTICES.md](docs/BEST-PRACTICES.md).

Measured results (reproducible)

Full AI trial walkthrough (in Chinese): [docs/TRIAL-REPORT.md](docs/TRIAL-REPORT.md) — 95-file mixed Rust/Python/TS repo, unknown symbol to def+callers in 3 calls ≈ 80 ms; honest limitations included.

Windows x64, 50 Rust files, 5 warmups, 30 samples (p95):