🇺🇸English

PST Email Extraction

Extract emails from Outlook PST files into an organised, integrity-verified archive of markdown files, raw email backups, and attachments. Supports full extraction and incremental append mode.

Prerequisites

• Python virtual environment set up (run setup.sh if not done)
• At least one of: libratom (Python) or readpst (system tool from pst-utils)

First-Time Setup

# Set up Python environment (one-time)
~/.claude/skills/pst-to-markdown/setup.sh

System Dependencies (optional fallback)

If libratom installation fails, install readpst as a fallback:

# Ubuntu/Debian
sudo apt install pst-utils

# macOS
brew install libpst

Extraction Operations

Full Extraction

Extract all emails from a PST file into markdown:

# Basic extraction
~/.claude/skills/pst-to-markdown/.venv/bin/python ~/.claude/skills/pst-to-markdown/scripts/extract_pst.py /path/to/file.pst /path/to/output/

# Verbose output with progress
~/.claude/skills/pst-to-markdown/.venv/bin/python ~/.claude/skills/pst-to-markdown/scripts/extract_pst.py /path/to/file.pst /path/to/output/ --verbose

# Include deleted items
~/.claude/skills/pst-to-markdown/.venv/bin/python ~/.claude/skills/pst-to-markdown/scripts/extract_pst.py /path/to/file.pst /path/to/output/ --include-deleted --verbose

# Set timezone for date display
~/.claude/skills/pst-to-markdown/.venv/bin/python ~/.claude/skills/pst-to-markdown/scripts/extract_pst.py /path/to/file.pst /path/to/output/ --timezone "Europe/London"

# Fix MAILER-DAEMON sent items (provide the PST owner's email)
~/.claude/skills/pst-to-markdown/.venv/bin/python ~/.claude/skills/pst-to-markdown/scripts/extract_pst.py /path/to/file.pst /path/to/output/ --owner-email "user@example.com"

Incremental Extraction (Append Mode)

Add only new emails (skips already-extracted messages by Message-ID):

~/.claude/skills/pst-to-markdown/.venv/bin/python ~/.claude/skills/pst-to-markdown/scripts/extract_pst.py /path/to/file.pst /path/to/output/ --append --verbose

Extract from Pre-Extracted .eml Directory

If emails were already extracted with readpst elsewhere, point at the directory:

~/.claude/skills/pst-to-markdown/.venv/bin/python ~/.claude/skills/pst-to-markdown/scripts/extract_pst.py /path/to/eml-directory/ /path/to/output/

Output Structure

output/
├── emails/
│   ├── FolderName/
│   │   ├── 2023-01-15_093042_from-john.smith_to-jane.doe_RE-Subject/
│   │   │   ├── email.md              # Formatted markdown with YAML frontmatter
│   │   │   ├── email.eml             # Raw original email (RFC 822)
│   │   │   ├── attachment_001_doc.pdf # Extracted attachments
│   │   │   └── checksums.sha256      # Per-email integrity hashes
│   │   └── .../
│   └── .../
├── index.csv                          # Machine-readable master index
├── index.md                           # Human-readable index with timeline
├── extraction_log.txt                 # Processing log with statistics
└── manifest.sha256                    # Master integrity manifest

Email Markdown Format

Each email.md contains:

• YAML frontmatter : message_id, date, from, to, cc, subject, attachments with SHA256 hashes
• Formatted body : HTML converted to markdown, or plain text preserved
• Attachment links : Relative links to extracted files with sizes
• Original headers : Full RFC 822 headers in code block

Index Files

• index.csv : All emails with date, sender, recipient, subject, folder, attachment count
• index.md : Timeline view grouped by year/month with links to each email

CLI Reference

extract_pst.py [-h] [--include-deleted] [--timezone TZ] [--verbose] [--append] [--owner-email EMAIL] pst_file output_dir

Extraction Backends

The tool tries backends in priority order:

1. libratom (Python) — preferred, installed via requirements.txt
2. readpst (system CLI) — fallback, from pst-utils package
3. Directory mode — processes pre-extracted .eml files directly

Integrity Verification

Every extraction produces a verifiable chain of custody:

1. Each email folder has checksums.sha256 (SHA256 of all its files)
2. manifest.sha256 hashes all checksum files plus the index
3. Source PST SHA256 is recorded in the manifest

To verify: sha256sum -c manifest.sha256

Workflow: Extract and Search

For large archives, combine with the repo-search skill:

1. Extract PST to markdown with this skill
2. Index the output with repo-search ingest
3. Semantic search across all emails

# Step 1: Extract

~/.claude/skills/pst-to-markdown/.venv/bin/python ~/.claude/skills/pst-to-markdown/scripts/extract_pst.py archive.pst ./email-output/ --verbose

# Step 2: Index for search (uses repo-search skill)
~/.claude/skills/repo-search/.venv/bin/python ~/.claude/skills/repo-search/ingest.py ./email-output/ --verbose

# Step 3: Search
~/.claude/skills/repo-search/.venv/bin/python ~/.claude/skills/repo-search/query.py --db-path ./email-output/.vectordb search "settlement agreement"

Error Handling

• "readpst not found" : Install pst-utils or ensure libratom is installed via setup.sh
• Corrupt emails : Logged to extraction_log.txt, processing continues
• Encoding issues : Falls back through UTF-8 → latin-1 → raw bytes
• Duplicate timestamps : Appended with -001, -002 suffixes
• Path too long : Subject truncated, uniqueness preserved

Performance

A typical 300MB PST (~1,000-3,000 emails) processes in 5-15 minutes.

Weekly Installs

1

Repository

dandcg/claude-skills

First Seen

1 day ago

Security Audits

Gen Agent Trust HubPassSnykPass

Installed on

amp1

cline1

opencode1

cursor1

kimi-cli1

codex1