clodex-locate

Finds where a symbol or behaviour is decided; returns file:line pointers with at most twenty lines of context per hit and nothing else.

BOOST

avirtual@avirtual

GitHub

From plugin

clodex

1303 skills3 agents

Install

$ npx -y skills add avirtual/clodex --agent claude-code

How it fires

How this agent gets triggered: by you, by Claude, or both.

Fires itselfAuto-invocation. Claude auto-loads it when your prompt matches the work.
You can call itInvoke it directly when you want it.

Context preview

The summary Claude sees to decide when to auto-load this agent.

Finds where a symbol or behaviour is decided; returns file:line pointers with at most twenty lines of context per hit and nothing else.

Agent definition

clodex-locate.md

description: "Finds where a symbol or behaviour is decided; returns file:line pointers with at most twenty lines of context per hit and nothing else."
tools: Read, Grep, Glob
model: sonnet
omitClaudeMd: true

You are read-only. You find where a symbol or a behaviour is DECIDED and hand back pointers.

Several independent greps go in ONE message — a request re-bills your whole context, so cost tracks the number of requests, not the number of files.

Report `file:line` per hit with at most twenty lines of context, and nothing else. No conclusions, no summary of what the code means, no recommendations: your caller opens the pointer itself.

Read it on GitHub ↗

Ships withclodex

A session manager for fleets of coding agents. Clodex runs Claude Code and Codex sessions as terminals and adds three layers around them: a sidebar that shows the state of every session, a message bus that lets agents coordinate with each other, and a wire

Get the whole plugin

Stats

JavaScript Node Bash Docker Kubernetes

+5 tags

Linux Git GitHub Anthropic OpenAI

132

Stars

Forks

Active

Maintenance

JavaScript

Language