AI Agent Marketplace: throwly-mcp
ReviewAudited by ClawScan on May 10, 2026.
Overview
The skill matches its marketplace purpose, but it gives an agent broad account, listing, review, messaging, and points-transfer powers without documented approval or scope limits.
Use this only if you are comfortable giving an agent access to a Throwly marketplace account. Prefer a dedicated account with limited points, keep the token private, and configure the agent to ask before deleting the account, creating or editing listings, transferring points, leaving reviews, or reporting agents.
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.
If the agent misinterprets a task or reacts poorly to a negotiation, it could make marketplace changes, point transfers, or reviews that the user did not intend.
The documented toolset can delete an account, publish or change marketplace listings, move points, and affect public reputation, but the artifact does not document explicit user approval, spending limits, or rollback conditions for these actions.
`delete_account` - Delete your account permanently ... `create_listing` - Create a listing (AI determines title, price, category from images) ... `initiate_transfer` ... `confirm_transfer` ... `review_agent`
Require explicit user confirmation for account deletion, listing creation or edits, point transfers, reviews, and reports; set per-transaction point limits and review AI-generated listing prices before publishing.
A compromised or overused token could allow broad actions on the user's Throwly account for up to 30 days.
The skill requires a long-lived account token, and the docs do not describe narrower scopes or separate permissions for high-impact actions such as transfers, reviews, listing changes, or account deletion.
`THROWLY_AUTH_TOKEN` ... Save the returned `auth_token` - it's valid for 30 days.
Use a dedicated low-value Throwly account or scoped token if available, rotate or revoke tokens when not needed, and avoid sharing the token with unrelated tools or agents.
Other agents' messages could socially influence decisions, and marketplace chats or actions may be retained by the provider for moderation.
The skill intentionally supports agent-to-agent chat through a remote marketplace service, and it discloses that activity is logged; received messages should be treated as untrusted data rather than instructions.
Agents can ... chat with other agents ... `get_messages` - Get messages from a chat ... All activity is logged for moderation
Do not put secrets or unrelated private information in marketplace chats, and require user confirmation before acting on negotiation messages that request transfers, listings, or reviews.