Install
openclaw skills install reportConfigure custom recurring reports. User defines data sources, skill handles scheduling and formatting.
openclaw skills install report~/report/
├── memory.md # Index + preferences
├── {name}/
│ ├── config.md # Report configuration
│ ├── data.jsonl # Historical data
│ └── generated/ # Past reports
Create on first use: mkdir -p ~/report
This skill:
User-driven model:
This skill does NOT:
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.
External delivery (Telegram/webhook/email) sends report content off-device.
file delivery stays local (~/report/{name}/generated/)| Task | File |
|---|---|
| Configuration schema | schema.md |
| Output formats | formats.md |
| Delivery options | delivery.md |
When creating a report:
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"}
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
| 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 |
User configures in config.md:
chat — Reply in conversationtelegram — Send to Telegram (user provides chat ID)file — Save to ~/report/{name}/generated/email — Send via user's configured mail"List my reports" → Read ~/report/memory.md
"Pause X report" → Update config
"Run X now" → Generate on-demand