Agent use cases
Cited playbooks: what an agent can do with the servers in the directory.
142 use cases
Every playbook cites a vendor page and was reviewed before publishing.
| Use case | Category | Effect | Open |
|---|---|---|---|
Create a read-only action plan for a websiteThe agent generates a grounded, read-only plan for a specific request against a public website, including evidence, risks, and confirmation requirements. | Operations and infrastructure | Read-only | |
Create automated surveillance rules via natural languageThe agent configures specific detection rules on the surveillance server based on the user's verbal instructions, such as notifying the user when a specific type of vehicle appears or when a person is detected at night. | Operations and infrastructure | Changes data in your accounts | |
Create cryptographically verifiable message receiptsA tamper-proof receipt is generated for all messages in a thread, providing a root hash that can be anchored on-chain for long-term proof of existence and authorship. | Operations and infrastructure | Changes data in your accounts | |
Debug failed webhook deliveriesDiagnoses why webhook deliveries failed by analyzing headers, status codes, and payloads to identify integration issues. | Operations and infrastructure | Read-only | |
Diagnose agent installation healthA comprehensive health report identifying configuration errors, connectivity issues, or security interference with the AI Commander agent on the target machine. | Operations and infrastructure | Read-only | |
Enforce per-session spend limits in OpenClawPrevents excessive API costs by blocking tool calls and model turns when a session's estimated usage exceeds a defined ceiling, ensuring predictable spending per session. | Operations and infrastructure | Read-only | |
Establish secure bidirectional communication channelsTwo parties establish a secure, ephemeral conversation thread with cryptographic proof of authorship and expiration, enabling timed interactions without shared secrets. | Operations and infrastructure | Changes data in your accounts | |
Execute Agent Onboarding and Earn Testnet USDCThe agent registers itself, completes a read-only tour of the platform, and receives 0.01 testnet USDC in an auto-provisioned wallet without human intervention. | Operations and infrastructure | Moves money or places orders | |
Execute approved physical access runbookA human operator performs a physical action such as a power cycle or cable check based on an approved runbook, returning status updates and evidence while keeping private keys secure. | Operations and infrastructure | Changes data in your accounts | |
Execute paid API calls via MCPThe agent successfully authenticates, purchases credits, and executes a paid API call through the APIHub proxy, receiving the upstream JSON response and payment confirmation headers. | Operations and infrastructure | Changes data in your accounts | |
Execute remote commands via AI agentThe AI agent executes shell commands on remote machines (VMs, workstations, or servers) by connecting through the MCP server, enabling automated system administration and remote task execution. | Operations and infrastructure | Changes data in your accounts | |
Execute supervised API write with approvalThe agent submits an intent, which is evaluated against deterministic policies and requires owner approval before credentials are brokered to execute the write, resulting in a recorded receipt of the action. | Operations and infrastructure | Changes data in your accounts |
Browse by category
Where these playbooks come from
Each playbook is drawn from the vendor page it cites and reviewed by Koskamo before it is published. Koskamo does not run these servers or your agent: check what each tool will do before you let it act on your accounts.
Koskamo reads the example, tutorial, and guide pages that vendors publish for their own MCP servers. A playbook is only drafted when the page says the server can do it, and the supporting quote is shown on every playbook.