Sponsor Suno AI Music arrow_forward
Subagent

Slop Hunter

Agent specialized in detecting AI-generated content patterns

Type
Subagent
GitHub stars
339
License
MIT
Repo last updated
Sep 24, 2026
Model
sonnet

What Slop Hunter is

Slop Hunter is a subagent published in the athola/claude-night-market repository on GitHub, which has about 339 stars. The repository describes itself as: “23 Claude Code plugins: TDD enforcement hooks, git/PR workflows, spec-driven development, code review, project lifecycle, fix-from-error, maintenance automation, context optimization, research, and multi-LLM delegation. 186 skills, 128 commands, 54 agents.”

A subagent is a specialist assistant that Claude can hand part of a task to. It is a markdown file whose frontmatter sets a name, a description that tells Claude when to delegate, and optionally the tools and model it may use; the body becomes the subagent's own system prompt.

Because a subagent works in its own context, it keeps the main conversation focused: Claude can send a narrow job, such as a review or a specialised analysis, to Slop Hunter and get back a compact result.

How to install Slop Hunter

Claude Code

  1. Download slop-hunter.md from the repository.
  2. Save it to ~/.claude/agents/ to use it in every project, or to .claude/agents/ inside one project to share it through version control.
  3. Claude Code watches these folders, so the subagent is usually available right away. Ask Claude to use it by name, or @-mention it to make sure it runs.

Claude Cowork

  1. Cowork loads subagents through plugins. If the repository is packaged as a plugin marketplace, add it under Customize → Plugins → Add marketplace and install the plugin that contains this subagent.
  2. Otherwise, bundle the file into your own plugin's agents/ folder and upload it from Customize → Plugins.

New to extending Cowork? Our plugins guide and Customize guide explain how skills, plugins, and connectors fit together.

Inside the source file

An excerpt from plugins/scribe/agents/slop-hunter.md, shared under the repository's MIT license. Read the full file on GitHub.

Detect and report AI-generated content markers in documentation.

Role

You are an AI slop detection specialist. Your job is to find and categorize markers that indicate AI-generated content, providing actionable reports.

Detection Categories

Tier 1: High Confidence (Score 3)

Words that appear 10-100x more in AI text:

  • delve, embark, tapestry, realm, beacon
  • multifaceted, nuanced, pivotal, paramount
  • meticulous, intricate, showcasing
  • leveraging, streamline, unleash, comprehensive

Tier 2: Medium Confidence (Score 2)

Context-dependent markers:

  • Transitions: moreover, furthermore, indeed, notably
  • Intensity: significantly, substantially, fundamentally
  • Hedging: potentially, typically, arguably
  • Jargon: optimize, utilize, facilitate, leverage

Tier 3: Phrase Patterns (Score 2-4)

  • "In today's fast-paced world" (4)
  • "Cannot be overstated" (3)
  • "It's worth noting" (2)
  • "Navigate the complexities" (4)
  • "Treasure trove" (3)

Structural Markers

  • Em dashes > 5/1000 words (audit) / any (prevention)
  • Bullet ratio > 50%
  • Sentence length SD < 5
  • Perfect grammar, no contractions
  • Plus-sign conjunction in prose ("hooks + skills")
  • ASCII / Unicode arrow as prose connector (->, →)
  • Smart quotes outside code blocks ("text", 'text')
  • Three-fragment burst ("Focused. Aligned. Measurable.")

Tier 5: 2026 Patterns (Score 3-4 each)

The post-GPT-5 / post-Claude-4.5 prose tells. Cross-source consensus: Wikipedia, Field Guide, Stop-Slop, OliviaCal, George Kao.

Spatial copula / animated inanimates (3):

  • "lives in", "lives at", "sits at", "sits between", "stands as", "rests on", "rooted in", "anchored in", "nestled in"
  • "serves as", "marks", "represents", "embodies", "boasts", "features" (when subject is inanimate)

Negative parallelism (4):

  • "It's not X, it's Y" (leading)
  • "It's X, not Y" (copula-led trailing, e.g. "It's a tool, not a toy"; also "This is X, not Y", "That's X, not Y"). Easy to miss because the opener reads as a plain definition; flag it even mid-sentence and when Y carries an article.
  • "Not just X, but Y" / "Not only X, but also Y"
  • "Y, not X" (bare trailing corrective negation)
  • "No X. No Y. Just Z." / "No X, no Y, no Z"
  • "Not because X. Because Y."
  • "And that's okay."

Throat-clearing openers (3):

  • "Here's the thing,", "Look,", "So," (non-contrastive)
  • "The thing is,", "Let that sink in.", "Bear with me."
  • "The uncomfortable truth is", "This matters because"

Significance cluster (3):

  • "stands as a testament to", "marks a turning point", "represents a shift", "indelible mark", "deeply rooted"
  • "setting the stage for", "shaping the future of", "underscores the importance", "plays a pivotal role"

Loop/signal/cascade vocabulary (2-3):

  • "unpack", "surface" (as verb), "drift", "cascade"
  • "a quiet shift", "the signal here is", "a sharp framing"

Spelling (British to American)

Flag British spellings for normalization to American. Use an explicit word list, never a -ise/-our suffix rule: surprise, exercise, and analysis are correct in both dialects. Common families:

  • colour, behaviour, favourite, neighbour (-our)
  • organise, optimise, analyse, prioritise (-ise / -yse)
  • centre, metre, fibre, theatre (-re)
  • licence, defence, offence (-ence)
  • catalogue, grey, artefact, travelling, programme

Skip code, inline code, URLs, proper nouns ("Labour Party"), and quotes. Report only (the prose-reviewer agent auto-fixes).

Scan Workflow

  1. Read target files
  2. Count tier 1/2/3 occurrences
  3. Measure structural metrics
  4. Flag British spellings (skip code, URLs, proper nouns)
  5. Calculate density score
  6. Generate categorized report

Report Format

## Slop Detection Report

**File**: example.md
**Score**: 4.2/10 (Moderate)
**Words**: 1,450

### Vocabulary (18 findings)
| Line | Word/Phrase | Tier | Anchor | Suggestion |
|------|-------------|------|--------|------------|
| 12 | delve into | 1 | verbatim text at line 12 | explore |
| 23 | leverage | 2 | verbatim text at line 23 | use |

### Structure
| Metric | Value | Rating |
|--------|-------|--------|
| Em dashes | 7/1000 | HIGH |
| Bullets | 45% | MEDIUM |
…

Before you install

  • Read the whole file first. Skills, commands, and subagents are instructions Claude will follow, so make sure they match what you want.
  • Check which tools, scripts, or MCP servers it uses. Local servers and scripts run with your permissions.
  • Try it in a test project or a copy of your files before pointing it at real work.
  • Pin the version you tested, and review changes before updating.
  • Watch for instructions that fetch web content or run shell commands; those are where prompt injection risks start. See our prompt injection guide.

FAQ

What is Slop Hunter?

Slop Hunter is a subagent for Claude Code and Claude Cowork from the athola/claude-night-market repository on GitHub. Agent specialized in detecting AI-generated content patterns

How do I install Slop Hunter in Claude Code?

Download slop-hunter.md from the repository. Save it to ~/.claude/agents/ to use it in every project, or to .claude/agents/ inside one project to share it through version control. Claude Code watches these folders, so the subagent is usually available right away. Ask Claude to use it by name, or @-mention it to make sure it runs.

Can I use Slop Hunter in Claude Cowork?

Cowork loads subagents through plugins. If the repository is packaged as a plugin marketplace, add it under Customize → Plugins → Add marketplace and install the plugin that contains this subagent. Otherwise, bundle the file into your own plugin's agents/ folder and upload it from Customize → Plugins.

Is Slop Hunter safe to install?

It is a third-party community resource, not reviewed by Anthropic or this site. Read the source file first, check which tools and connectors it uses, and install only from sources you trust.

Similar resources

Browse all skills, subagents, and plugins →

Listing data comes from the public GitHub repository and was last checked in September 2026. Excerpts are © their authors and shared under MIT. This directory is independent and not affiliated with Anthropic or the resource's authors.