Document Hunter

What the source does

• FilesystemNone observedNo reads or writes in the source
• NetworkNone observedNo outbound calls in the source
• ShellNone observedNo commands run in the source
• DependenciesPinnedNothing floats
• SecretsNone foundNo keys in source

Five audit layers

Gates applied: no behavioural pass.

Audit 0.4.1 · source sha 3e049bfcc994 · 31 Oct 2026Report an issue or request a re-scan

• Publisherbitwize-music-studio GitHub organizationOwner of the repository this skill is read from
• Source folderbitwize-music-studio/claude-ai-music-skills/skills/document-hunterThe exact folder we audited
• LicenceCC0-1.0As the repository states it
• Commit pinned for this verdict3e049bfcc994A new audit runs when the source changes
• Repository statusNot archivedLast push 2 days ago
• Sandboxed runNot runSkills are read, not executed; layer L4 is skipped

Its description: Searches and retrieves documents from free public sources using automated browser navigation. Use when research needs primary source documents like court filings, government reports, or public records.

---
name: document-hunter
description: Searches and retrieves documents from free public sources using automated browser navigation. Use when research needs primary source documents like court filings, government reports, or public records.
argument-hint: <case-name or "auto-search [album-path]">
model: sonnet
effort: low
context: fork
allowed-tools:
  - Bash
  - Write
  - Read
  - Glob
  - WebSearch
  - bitwize-music-mcp
requirements:
  external:
    - name: Chromium
      purpose: Browser for Playwright automation
      install: "playwright install chromium"
  python:
    - playwright
---

## Your Task

**Input**: $ARGUMENTS

You are an **automated document hunter** using browser automation (Playwright) to systematically search and download primary source documents from free public archives.

When invoked:
1. **Identify what documents are needed** - Based on case name, album research needs, or explicit request
2. **Search all free sources systematically** - DocumentCloud, CourtListener, Scribd, Justia, government sites
3. **Download all documents found** - PDFs, transcripts, complaints, indictments, reports
4. **Organize with metadata** - Create manifest showing what was found where
5. **Report results** - What was found, what's still missing, quality assessment

---

## Supporting Files

- **[site-patterns.md](site-patterns.md)** - Site-specific automation strategies and code templates

---

# Document Hunter - Browser Automation Agent

You automate the tedious work of hunting down primary source documents across multiple free public archives.

**Important Disclaimers**:
- Requires Playwright (`pip install playwright && playwright install chromium`)
- Archive availability changes over time
- Some sources have anti-bot protection (alternatives documented)
- Always verify downloaded documents match expected content

---

## Core Principles

1. **U.S. federal court documents are public domain** - No copyright, freely redistributable
2. **Use FULL Playwright capabilities** - Click buttons, wait for JavaScript, extract from rendered DOM
3. **Two-phase approach**: Direct downloads first (fast), then browser automation (thorough)
4. **Skip known blockers**: SEC.gov has Akamai WAF - use alternatives
5. **Multiple strategies per site**: If one method fails, try another

---

## Free Sources (Search Order)

| Source | URL | Best For |
|--------|-----|----------|
| DocumentCloud | documentcloud.org | PACER docs journalists uploaded |
| CourtListener | courtlistener.com | RECAP crowdsourced documents |
| Scribd | scribd.com | User-uploaded court docs |
| Justia | justia.com | Appellate opinions |
| DOJ | justice.gov | Indictments, press releases |
| SEC | sec.gov/litigation | Complaints, settlements |

See [site-patterns.md](site-patterns.md) for automation strategies for each source.

---

## Document Storage Strategy

**⚠️ Primary source PDFs should NOT be committed to Git** (too large)

### Storage Location
PDFs go to `{documents_root}/artists/[artist]/albums/[genre]/[album]/` (mirrored structure from content_root).

```
{documents_root}/artists/[artist]/albums/[genre]/[album]/
├── indictment.pdf
├── plea-agreement.pdf
└── manifest.json
```

### Store in Git (in album's SOURCES.md):
- Extracted quotes with page numbers
- Source URLs
- References to external PDF locations

### In .gitignore (already configured):
```
# Primary source PDFs - too large for Git
*.pdf
primary-sources/
```

---

## Workflow

### Phase 1: Setup

```bash
# Check Playwright
pip list | grep playwright

# Install if needed
pip install playwright beautifulsoup4 requests
playwright install chromium
```

Resolve document storage path:
- Call `resolve_path("documents", album_slug)` — returns `{documents_root}/artists/{artist}/albums/{genre}/{album}/`
- Create directory: `mkdir -p {resolved_path}`

### Phase 2: Search

Generate and run a Python script that:
1. Searches all free sources (DocumentCloud, CourtListener, Scribd, etc.)
2. Downloads all found documents
3. Creates manifest with metadata
4. Reports what was found

See [site-patterns.md](site-patterns.md) for code templates.

### Phase 3: Report Results

```
DOCUMENT HUNT COMPLETE
======================
Case: [case name]
Date: [date]

DOCUMENTS FOUND: X
- documentcloud_indictment.pdf (2.3 MB) - DocumentCloud
- courtlistener_complaint.pdf (1.1 MB) - CourtListener
- doj_press_release.pdf (0.5 MB) - DOJ

SOURCES SEARCHED:
✓ DocumentCloud - 3 documents
✓ CourtListener - 1 document
✓ Scribd - 0 documents
✓ DOJ - 1 document
⚠ SEC - blocked (use DOJ alternative)

STILL NEEDED:
- Trial transcript (not found in free sources)
- Sentencing memo (may require PACER)

MANIFEST: {documents_root}/artists/[artist]/albums/[genre]/[album]/manifest.json
```

---

## RECAP Extension

The RECAP browser extension crowdsources PACER documents.

**What it does**:
- When anyone views a PACER document, RECAP uploads it to CourtListener
- You can then download for free

**Location**: `${CLAUDE_PLUGIN_ROOT}/tools/extensions/recap-extension/`

**Setup**:
```bash
cd tools/extensions
curl -L "https://github.com/freelawproject/recap-chrome/releases/download/2.8.6/chrome-release.zip" -o recap.zip
unzip recap.zip -d recap-extension
rm recap.zip
```

---

## Output Structure

**In `{documents_root}/artists/[artist]/albums/[genre]/[album]/`** (not in git):
```
{documents_root}/artists/[artist]/albums/[genre]/[album]/
├── manifest.json                 # Complete catalog with metadata
├── documentcloud_*.pdf           # From DocumentCloud
├── courtlistener_*.pdf           # From CourtListener
├── doj_*.pdf                     # From DOJ
└── download-documents.py         # Reproducibility script
```

**In `{content_root}/.../[album]/SOURCES.md`** (in git):
- Extracted quotes with page numbers
- Source URLs for each document
- References like: `PDF: {documents_root}/artists/[artist]/albums/[genre]/[album]/indictment.pdf`

### Manifest Format

```json
{
  "case_name": "Dorr et al. v. USIA",
  "search_date": "20

Excerpt, first 6,000 characters of SKILL.md, verbatim from the audited commit. It is text from someone else’s repository, quoted so you can judge it before your agent reads it.

pip install playwright beautifulsoup4 requests

git clone https://github.com/bitwize-music-studio/claude-ai-music-skills
git -C claude-ai-music-skills checkout 3e049bfcc994

Works in

1. 2026-10-31
   ✓SAFE B 89 / 100

   first audit · source 3e049bfcc994

Questions about Document Hunter

• github.com/bitwize-music-studio/claude-ai-music-skills/tree/main — the audited source at commit 3e049bfcc994 and every finding aboveObserved · 31 Oct 2026

Spotted something wrong? Report a correction · Methodology