report

Data Storage

~/report/
├── memory.md               # Index + preferences
├── {name}/
│   ├── config.md           # Report configuration
│   ├── data.jsonl          # Historical data
│   └── generated/          # Past reports

Create on first use: mkdir -p ~/report

Scope

This skill: - ✅ Stores report configurations in ~/report/ - ✅ Generates reports on schedule - ✅ Delivers via channels user configures

User-driven model: - User defines WHAT data to include - User grants access to any needed sources - User provides API keys if external data needed - Skill handles SCHEDULING and FORMATTING

This skill does NOT: - ❌ Access APIs without user-provided credentials - ❌ Pull data from sources user hasn't specified - ❌ Store credentials (user provides via environment)

Environment Variables

No fixed requirements. User provides API keys as needed:

# Example: if user wants Stripe data
export STRIPE_API_KEY="sk_..."

# Example: if user wants GitHub data  
export GITHUB_TOKEN="ghp_..."

Config references env var name, never the value.

Delivery Security

External delivery (Telegram/webhook/email) sends report content off-device. - User explicitly configures each channel - User responsible for trusting destination - file delivery stays local (~/report/{name}/generated/)

Quick Reference

Task	File
Configuration schema	`schema.md`
Output formats	`formats.md`
Delivery options	`delivery.md`

Core Rules

1. User Defines Data Sources

When creating a report: 1. User specifies what data to track 2. If external API needed, user provides credentials 3. Credentials stored as env var references, not values

Example:

User: "Weekly report on my Stripe revenue"
Agent: "I'll need Stripe API access. Please set 
        STRIPE_API_KEY in your environment."
User: "Done"
→ Config stored with "source": {"type": "api", "env": "STRIPE_API_KEY"}

2. Report Configuration

In ~/report/{name}/config.md:

name: weekly-revenue
schedule: "0 9 * * 1"  # Monday 9am
sources:
  - type: api
    env: STRIPE_API_KEY  # User provides
format: chat
delivery: telegram

3. Scheduling

Frequency	Cron	Example
Daily	`0 9 * * *`	9am daily
Weekly	`0 9 * * 1`	Monday 9am
Monthly	`0 9 1 * *`	1st of month
On-demand	-	When user asks

4. Delivery Channels

User configures in config.md: - chat — Reply in conversation - telegram — Send to Telegram (user provides chat ID) - file — Save to ~/report/{name}/generated/ - email — Send via user's configured mail

5. Managing Reports

"List my reports" → Read ~/report/memory.md
"Pause X report" → Update config
"Run X now" → Generate on-demand