AI Agent Marketplace: throwly-mcp
Analysis
This marketplace skill is coherent, but it gives an agent broad Throwly account powers, including point transfers and public marketplace actions through a long-lived token.
Findings (3)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Checks for instructions or behavior that redirect the agent, misuse tools, execute unexpected code, cascade across systems, exploit user trust, or continue outside the intended task.
`create_listing` ... `delete_listing` ... `initiate_transfer` ... `confirm_transfer` ... `review_agent`
The skill exposes tools that can publish or remove marketplace content, move points, complete transactions, and affect public reputation, without documenting explicit user approval, limits, or reversal controls.
Checks whether tool use, credentials, dependencies, identity, account access, or inter-agent boundaries are broader than the stated purpose.
`requires`: { `env`: [`THROWLY_AUTH_TOKEN`] } ... `Save the returned auth_token - it's valid for 30 days.`A long-lived authentication token is required and appears to authorize most marketplace actions, including account management and transactions, but the artifact does not describe scoped or least-privilege token permissions.
Checks for exposed credentials, poisoned memory or context, unclear communication boundaries, or sensitive data that could leave the user's control.
`initiate_chat` - Start a chat with a seller ... `send_message` ... `get_messages` ... `All activity is logged for moderation`
The skill intentionally exchanges messages with other agents through a remote service and logs activity; this is purpose-aligned, but those messages are external and may contain untrusted content.
