setup-corpus

name	setup-corpus
description	First-run setup for SpamHole. Creates the user's spam-corpus workspace (configurable path) and seeds the JSON inventories — tracking-domains, sender-blocklist, processed-emails, ad-trackers. Use when the user first installs the plugin or wants to point it at a different corpus location.

setup-corpus

Scaffold a per-user SpamHole corpus — the directory tree that holds spam samples, blocklists, and tracking inventories. Every other skill in this plugin reads/writes into this corpus.

Resolving the corpus path

Resolve in this order:

1. If the user passed a path argument, use it.
2. If ${CLAUDE_USER_DATA:-$HOME/.local/share/claude}/spamhole/config.json exists and has corpus_path, use that.
3. Otherwise default to ${CLAUDE_USER_DATA:-$HOME/.local/share/claude}/spamhole/corpus/.

If the user wants the corpus inside a private GitHub repo (recommended — raw .eml may contain personal data), ask for the repo path and use that. Make sure the corpus directory is gitignored or that the user understands the privacy trade-off.

Steps

1. Confirm the resolved path with the user. Create it if it doesn't exist.

2. Build the layout:

   <corpus>/
   ├── inputs/
   │   └── emails/
   │       └── sanitised/        # Redacted samples safe to share
   ├── data/
   │   ├── tracking-domains.json # array — pixel + click-tracker hosts
   │   ├── ad-trackers.json      # array — adsense / DSP / analytics pixel hosts
   │   ├── sender-blocklist.json # array — blocked senders
   │   └── processed-emails.json # array — analysis log
   ├── outputs/
   │   └── analyses/             # per-email markdown reports
   └── exports/                  # generated DNS blocklists, filter rules, etc.
   

3. Initialize each JSON file as an empty array [] if it doesn't exist. Never overwrite an existing inventory.

4. Write <corpus>/README.md summarising what's in each directory and reminding the user this is their personal corpus.

5. Persist the resolved path to ${CLAUDE_USER_DATA:-$HOME/.local/share/claude}/spamhole/config.json so other skills can find it without re-prompting:

   { "corpus_path": "/abs/path/to/corpus" }
   

6. Suggest (do not require) committing the corpus to a private repo for backup, and offer to add a .gitignore that excludes raw .eml while tracking the JSON inventories and sanitised samples.

7. Report: corpus path, what was created vs already existed, suggested next step (capture-spam-sample).