# Hunt Information Disclosure

> Hunt error leakage, DVCS exposure, source maps, config files, and differential oracles.

- Skill: `uphiago/hunt-information-disclosure` (Agent Skill)
- Install (CLI): `npx skillmds@latest add uphiago/hunt-information-disclosure`
- Raw SKILL.md: https://api.skillmd.com/api/skills/uphiago/hunt-information-disclosure/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- License: MIT
- Author: uphiago (https://skillmd.com/u/uphiago)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/uphiago/hunt-information-disclosure

---


# Information Disclosure Hunting

Hunt for information exposure through stack traces, debug endpoints, versioned path discovery, source maps, and differential oracles. Each disclosure amplifies other vulnerabilities — a version number enables CVE targeting, a server path enables LFI, a schema leak enables auth bypass, and an error message reveals internal infrastructure.

## When to Use

- Applications return verbose error messages with stack traces, file paths, or SQL fragments.
- Source maps (.js.map) are deployed to production.
- Versioned static assets reveal framework/CMS versions.
- API responses differ by object existence (user enumeration by status/length/time).
- Debug endpoints, health checks, or status pages expose internal state.

## Quick Detection

```bash
# Trigger errors on common paths
for path in "/nonexistent" "/%00" "/.." "/error" "/debug"; do
  curl --max-time 30 --connect-timeout 10 -sk "https://target.com$path" | grep -iE "stack|trace|exception|error|warning|debug|line [0-9]+" | head -5
done
```

## Procedure

### Phase 1 — Error & Exception Leakage

```bash
# Trigger errors with malformed input
curl --max-time 30 --connect-timeout 10 -sk "https://target.com/api/users?id='"
curl --max-time 30 --connect-timeout 10 -sk -X POST "https://target.com/api/login" -d '{"username":null}'
curl --max-time 30 --connect-timeout 10 -sk "https://target.com/search?q=%00"

# Check response for sensitive data
# Stack traces → file paths, line numbers, framework version
# SQL errors → table names, column names, DB type
# Deserialization errors → class names, serialization format
# Template errors → template paths, engine type

# Fuzz for debug endpoints
for path in "/debug" "/__debug__" "/debugbar" "/_debug_toolbar" "/.well-known/debug" \
            "/actuator" "/actuator/info" "/actuator/env" "/actuator/health"; do
  curl --max-time 30 --connect-timeout 10 -sk "https://target.com$path" -w "\n%{http_code} — $path\n" -o /dev/null
done
```

### Phase 2 — DVCS & Config File Discovery

```bash
# Git, SVN, Mercurial exposure
for path in "/.git/HEAD" "/.git/config" "/.svn/entries" "/.hg/store/"; do
  curl --max-time 30 --connect-timeout 10 -sk "https://target.com$path" -w "%{http_code} — $path\n" -o /dev/null
done

# Config and env files
for pattern in ".env" ".env.local" ".env.production" ".env.staging" \
               "config.json" "config.yml" "settings.py" "settings.php" \
               "wp-config.php" "web.config" "app.config"; do
  curl --max-time 30 --connect-timeout 10 -sk "https://target.com/$pattern" -w "%{http_code} — $pattern\n" -o /dev/null
done

# Backup files
for pattern in "backup.zip" "backup.sql" "dump.sql" "db.sql" \
               "database.sql" "export.sql" "site.tar.gz" "backup.tar.gz"; do
  curl --max-time 30 --connect-timeout 10 -sk "https://target.com/$pattern" -w "%{http_code} — $pattern\n" -o /dev/null
done
```

### Phase 3 — Source Map Exploitation

```bash
# Find .js.map files
curl --max-time 30 --connect-timeout 10 -sk "https://target.com" | grep -Eo '[^"\s]+\.js\.map' | sort -u

# Download and extract
wget "https://target.com/static/app.js.map"
node -e "
const m=require('./app.js.map');
m.sources.forEach((s,i)=>require('fs').writeFileSync(s.split('/').pop(),m.sourcesContent[i]));
console.log('Extracted '+m.sources.length+' files');
"

# Look for NEXT_PUBLIC env vars in extracted source
grep -r "NEXT_PUBLIC_" extracted_files/ | cut -d= -f1 | sort -u
```

### Phase 4 — Differential Oracles

```bash
# User enumeration via status code
for id in {1..50}; do
  curl --max-time 30 --connect-timeout 10 -sk "https://target.com/api/users/$id" -w "$id — %{http_code}\n" -o /dev/null
done

# Object existence via response size
for id in {1..100}; do
  size=$(curl --max-time 30 --connect-timeout 10 -sk "https://target.com/api/orders/$id" -w "%{size_download}" -o /dev/null)
  echo "$id — $size bytes"
done | awk '$2 > 100 {print "EXISTS: "$0}'

# Timing oracle for blind enumeration
for id in {1..50}; do
  time curl --max-time 30 --connect-timeout 10 -sk "https://target.com/api/users/$id" -o /dev/null -w "%{time_total}s $id\n"
done | sort -rn

# ETag/304 oracle
for id in {1..10}; do
  etag=$(curl --max-time 30 --connect-timeout 10 -skI "https://target.com/api/users/$id" | grep -i etag)
  echo "$id: $etag"
done
```

### Phase 5 — Version & Technology Discovery

```bash
# Framework version from static assets
curl --max-time 30 --connect-timeout 10 -sk "https://target.com/static/admin/css/base.css" | head -5
curl --max-time 30 --connect-timeout 10 -sk "https://target.com" | grep -Eo '(?:Django|Laravel|Rails|Express|Next\.js|Nuxt)[\s/]*v?[0-9.]+'

# Package manager lock files
curl --max-time 30 --connect-timeout 10 -sk "https://target.com/composer.lock" | jq -r '.packages[] | select(.version) | "\(.name)@\(.version)"' 2>/dev/null
curl --max-time 30 --connect-timeout 10 -sk "https://target.com/package-lock.json" | jq -r '.packages | to_entries[] | "\(.key)@\(.value.version)"' 2>/dev/null
curl --max-time 30 --connect-timeout 10 -sk "https://target.com/yarn.lock" | head -30
```

### Phase 6 — Chaining Disclosures

Each disclosure type provides inputs for other attack vectors:

| Disclosure | Chains to |
|---|---|
| Framework version | CVE database lookup |
| Server path | LFI path traversal |
| Internal IP | SSRF target |
| API schema | Auth bypass via undocumented endpoint |
| Dependency version | Supply chain vulnerability |
| NEXT_PUBLIC variables | API key/Supabase/Firebase access |
| SQL error | SQL injection confirmation + DB type |

## Pitfalls

- **Not every error message is exploitable.** A generic "An error occurred" page with no details is not a finding.
- **Source maps may be empty or stripped.** Verify extracted content before reporting.
- **Differential oracles are statistical.** Confirm with at least 3 samples before reporting.
- **`.git` exposure must contain actual repo data, not just HTTP 200 on a path.** A catch-all SPA may return 200 for `/.git/HEAD` without serving git data.
- **Version disclosure alone is usually LOW severity.** Chain it — version → CVE → exploit.

## Verification

1. Error message contains actionable internal data (file path, SQL query, framework version).
2. Source map extraction produces real source files with identifiable code (not just webpack bootstrap).
3. Differential oracle consistently distinguishes between existing and non-existing resources.
4. Chain the disclosure to another vulnerability before reporting — standalone info disclosure is rarely critical.
5. For config files: the leaked data must contain credentials, API keys, or connection strings (not just generic config).

## Related Skills

- **`source-leak-hunt`** — Focused on `.env`, `.git`, and config file leakage.
- **`js-secrets-extraction`** — API keys and tokens in JavaScript bundles.
- **`error-log-mining`** — PHP error logs with credential and query leakage.
- **`web-enumeration`** — Path discovery that reveals sensitive endpoints.