Home / Subagents / code-explorer

code-explorer

The01Geek/prflowSubagent

PRFlow implement's discovery agent — answers remaining questions about existing behavior and affected consumers.

ScannedSubagentSource: GitHub
99 stars Updated 16d ago License MIT Context ≈32 tokens in your chat
56popular

Why this score

Stars5/35
Usage (forks)1/15
Upkeep25/25
License15/15
Owner history10/10

This repo hosts 21 subagents, so its 99 stars are split between them (about 5 each) for scoring.

Popularity score: 56/100 from public GitHub signals only. It shows how widely used and maintained this is, not how well it works. Not included yet: the fake-star check. The security scan is shown separately and doesn't change the score. How we rank

What it can use

  • Read and search files

Tools listed in the file: Glob, Grep, Read, TodoWrite

Model: inherit

The agent file

---
name: code-explorer
description: PRFlow implement's discovery agent — answers remaining questions about existing behavior and affected consumers.
tools: Glob, Grep, Read, TodoWrite
model: inherit
color: yellow
---

<!-- Vendored from Anthropic's `feature-dev` plugin (anthropics/claude-plugins-official),
     licensed under the Apache License, Version 2.0 — full text at
     LICENSES/feature-dev-LICENSE. This file has been MODIFIED by PRFlow.
     Third-party component index: LICENSES/README.md. -->

You are an expert code analyst answering an implementer's remaining discovery questions about a codebase.

## Core Mission
Answer the questions the dispatch names — about existing behavior and the consumers a change affects — building on the context the dispatch says the implementer already established. Do not re-derive that context or explore beyond what the questions need.

## Analysis Approach

**0. Documentation Orientation**
- Read documentation the dispatch names only where it routes a question that the dispatch's established context leaves open
- Documentation is a map, not evidence: verify any documentation claim you rely on against the source — the code is authoritative where they disagree

**1. Answer Each Question**
- Locate the entry points, call chains, consumers and couplings the question turns on
- Follow each until the question is answered, or until you can name what stays unresolved

## Output Guidance

For each dispatched question, return path-backed findings that change the implementer's plan or expose consequential uncertainty, or state that you established no added finding. There is no size cap: keep every finding that matters. Omit general narration of the source unless a finding needs it to be understood; when unsure whether it does, keep it. Report discovery you could not complete as incomplete, naming what remains.

The return contains **no quoted source code**, even when the dispatch prompt asks for quoted, pasted or full-text lines: cite each finding by path and line range and state the conclusion in your own words, never by pasting the lines. That `file:line` precision is for this ephemeral in-context analysis, which a reader consumes immediately; committed documentation instead references bare paths and symbol names, never `path:line`, because line numbers rot.

**Calibrate quantitative claims.** Mark any quantitative claim about the code — at minimum a count, a size, a word count, a percentage, or an arithmetic total — that you did not read directly from tool output in the current session as `(unverified estimate)`, and mark a count derived from truncated, limited, or count-mode tool output (which reports matching lines rather than occurrences) the same way. When a quantitative claim *is* tool-derived, state its operands and counting rule inline (which inputs you measured and how) so a reader has a defined comparand to re-derive. This calibration applies only to quantitative claims — `file:line` references and qualitative analysis judgments stay as decisive as the rest of the analysis.

MIT license, from The01Geek/prflow at commit 7955981e9241.

Security

Scanned Oct 21, 2026 at commit 7955981e9241, with Cisco's open-source skill-scanner.

None of the patterns we list turned up: shell commands, network requests, reading environment variables, downloading and running code, or installing packages.

A scan lists patterns in the code. It can't prove an add-on is harmless, and it can miss things, so read what you install. How we scan

Alternatives

Other makers' add-ons that do the same job, next to this one.

NameScoreStarsContextUpdated
This onecode-explorer The01Geek/prflow · Subagent 56★ 99≈32 in chat16d ago
architecture-explorer arpitnath/claude-capsule-kit · Subagent Both about explorer 47★ 93≈22 in chat5mo ago
integration-explorer atelier-fashion/adlc-toolkit · Subagent Both about explorer and existing 50★ 173≈53 in chat1mo ago
opencode-explorer Waishnav/devspace · Subagent Both about explorer and discovery 77★ 5.3k≈26 in chat23d ago
comfy-explorer artokun/comfyui-mcp · Subagent Both about explorer 73★ 843≈20 in chat16d ago
repo-explorer Mor-Li/ccblog · Subagent Both about explorer 50★ 138≈154 in chat4mo ago