Use Cases for OpenClaw

Pre-built workflows you can deploy to your OpenClaw instance

Each use case is a guided setup session that walks you through connecting external services, configuring DNS, collecting API keys, and deploying OpenClaw agents — all through the same tunnel you already use for setup.

Frequently Asked Questions

How does Zedly Shield protect my data during workflows?

Zedly Shield is a runtime security layer (shieldd daemon) that sits between your OpenClaw agents and the LLM providers. Before any content leaves your machine, Shield intercepts it and replaces personally identifiable information — email addresses, phone numbers, names, financial details — with opaque tokens like [EMAIL_1] or [PHONE_2]. The LLM only ever sees tokenized data. When the response comes back, Shield rehydrates the tokens with the original values locally. Your raw data never reaches third-party servers.

Is my email content sent to third-party AI providers?

Not in raw form. When workflows like Intelligent Email Triage process your inbox, Zedly Shield automatically redacts all PII before the content is sent to an LLM for classification or reply drafting. The LLM sees sanitized text with placeholder tokens. After processing, Shield restores the original values locally on your machine. You can verify and adjust exactly what gets redacted through the Shield dashboard at /shield on your Zedly instance.

Can I customize what Zedly Shield redacts?

Yes. Shield policies are fully configurable through the /shield dashboard or directly in the shieldd.toml configuration file. You can enable or disable redaction categories (emails, phone numbers, addresses, financial data), add specific terms to an allowlist so they pass through unredacted, or create custom redaction patterns for domain-specific sensitive data like internal project codes or client IDs.

Are my API keys and credentials stored securely?

Yes. Credentials you provide during setup — Gmail OAuth tokens, IMAP passwords, Telegram bot tokens — are encrypted and stored exclusively on your machine inside the OpenClaw credential store. Zedly’s servers never see or store your API keys. The setup tunnel only carries commands and responses; credential values are written directly to your local filesystem through the SSH connection.

What happens to my data after a setup session ends?

When you disconnect the SSH tunnel, the session key is automatically revoked and no further access is possible. Zedly retains a log of commands executed during the session for your audit trail, but no files, emails, documents, or credentials are copied from your machine. All workflow data — processed emails, draft replies, memory files — stays entirely on your hardware.

Do workflows work with any email provider?

The email workflows support Gmail natively via OAuth 2.0 with granular API scopes, and any other provider (Outlook, Yahoo, Fastmail, self-hosted) via standard IMAP/SMTP with app passwords. Gmail users get richer features like label management and thread grouping. IMAP users get full inbox scanning, classification, and reply drafting with the same Shield protection.

Does the AI agent learn and remember over time?

Yes. Workflows that support learning — like Intelligent Email Triage — use OpenClaw’s three-layer memory architecture. Daily Notes capture short-term context, QMD (Quiescent Memory Distillation) provides medium-term retrieval using hybrid BM25 and vector search, and Dreaming consolidates long-term patterns overnight. The agent gets smarter at categorizing your email, matching your writing style, and recognizing your VIP contacts over weeks and months.

Is each agent’s data isolated from other agents?

Yes. Every OpenClaw agent runs in its own workspace with a separate memory store, configuration files, and credential scope. A triage agent cannot read a cold-outreach agent’s emails or memory, and vice versa. QMD indexes and Dreaming consolidation are scoped per-agent, so learned patterns from one workflow never leak into another.

Start Use Case

Select one of your connected sessions to run this workflow on.

Prerequisites

    Loading sessions...