cortex-mine
Mine Gmail history into a local flat-file knowledge base (~/.cortex/). Use when asked to "run the cortex", "mine emails", "cortex run", "cortex dry run", "set up the cortex", "cortex from DATE", or "mine my inbox". Extracts contacts, clients, communications and knowledge facts into portable JSONL/JSON files. Requires gws CLI and ANTHROPIC_API_KEY.
How do I install this agent skill?
npx skills add https://github.com/jezweb/claude-skills --skill cortex-mineIs this agent skill safe to install?
- Gen Agent Trust Hubpass
This skill mines Gmail history into local files using the Anthropic API. It requires local installation of the Google Workspace CLI and the Anthropic Python library. Security considerations include the processing of private email data and the potential for indirect prompt injection from email content.
- Socketpass
No alerts
- Snykfail
Risk: HIGH · 2 issues
- Runlayerfail
5/5 files flagged
What does this agent skill do?
Knowledge Cortex — Mine
Mine your Gmail history into structured flat files — contacts, clients, communications, and key facts — stored in ~/.cortex/. No server required. Files you own.
When invoked
Read {skill_path}/references/file-schema.md for the storage format before any file operations.
Commands
| What user says | Action |
|---|---|
| "run the cortex" / "cortex run" | Run mining pass |
| "cortex dry run" | Mine without writing — pass --dry-run |
| "cortex from DATE" | Mine from specific date — pass --from YYYY-MM-DD |
| "set up the cortex" | Run first-time setup |
Setup (first run)
Check prerequisites:
which gws || echo "MISSING: npm install -g @googleworkspace/cli"
python3 -c "import anthropic" 2>/dev/null || echo "MISSING: pip install anthropic"
ls ~/.cortex/state.json 2>/dev/null || echo "FIRST RUN"
If gws missing: tell user npm install -g @googleworkspace/cli then gws auth setup.
If anthropic missing: tell user pip install anthropic (or pip install anthropic --break-system-packages on managed Python).
If first run: initialise storage.
mkdir -p ~/.cortex/originals
echo '{}' > ~/.cortex/contacts.json
echo '{}' > ~/.cortex/clients.json
touch ~/.cortex/communications.jsonl ~/.cortex/knowledge.jsonl ~/.cortex/files.jsonl
Write ~/.cortex/state.json:
{"version":"1.0","cursors":{"gmail":null},"last_run":null,"totals":{"contacts":0,"clients":0,"communications":0,"knowledge":0,"files":0},"runs":[]}
Mining
python3 {skill_path}/scripts/cortex-mine.py
python3 {skill_path}/scripts/cortex-mine.py --dry-run
python3 {skill_path}/scripts/cortex-mine.py --from 2024-01-01
python3 {skill_path}/scripts/cortex-mine.py --batch-size 25
The script prints a run report on completion. Show it to the user verbatim.
Common errors:
gws: command not found— install gws CLIAuthError/403— rungws auth setupANTHROPIC_API_KEY not set— set env var or add to~/.cortex/.envModuleNotFoundError: anthropic—pip install anthropic
Scheduling
- Claude Code Desktop: Cowork > Scheduled > New Task >
Run the cortex-mine skill> Daily - CLI:
/loop 24h run the cortex-mine skill - Cron:
0 6 * * * cd ~ && ANTHROPIC_API_KEY=sk-... python3 /path/to/cortex-mine.py
Exporting data
See {skill_path}/references/export-adapters.md for patterns to export cortex data to Obsidian, SQLite, Notion, CRM systems, or custom APIs.
Environment
| Variable | Default | Purpose |
|---|---|---|
CORTEX_DIR | ~/.cortex | Storage root |
ANTHROPIC_API_KEY | — | Required for extraction |
CORTEX_BATCH_SIZE | 50 | Threads per run |
CORTEX_OWNER_EMAIL | — | Your email — excluded from contacts |
References
- File schemas:
{skill_path}/references/file-schema.md - Pre-filter rules:
{skill_path}/references/prefilter-patterns.md - Export patterns:
{skill_path}/references/export-adapters.md
How can the creator link this skill?
Add the canonical catalog link to the repository README so users can inspect current installs and available audits. The publishing guide covers the complete discovery path.
<a href="https://skillzs.dev/skills/jezweb/claude-skills/cortex-mine">View cortex-mine on skillZs</a>