Seisoai
ReviewAudited by ClawScan on May 10, 2026.
Overview
Seisoai is a coherent media-generation gateway, but it can trigger paid API or x402 requests without a mandatory price check, budget, or user approval step.
Install this only if you want your agent to call Seisoai’s remote media-generation APIs. Set spending limits, require confirmation before paid jobs, verify pricing and tool schemas before each request, and use dedicated low-privilege credentials.
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 has usable credentials or a payment-capable wallet, it could consume credits or spend USDC on media-generation requests without a clear confirmation step.
The skill supports paid x402/API-key invocations but does not require a price check, budget limit, or explicit user approval before paid media jobs.
"Optionally pre-price inputs" ... "Invoke with API key or x402" ... "normal media generation ... no extra checks"
Require explicit user approval and a maximum budget before any paid invocation; make pre-pricing mandatory; prefer API keys with quotas or spending caps.
A compromised or over-broad API key could expose the user’s Seisoai account usage or allow unwanted paid requests.
The skill uses provider credentials or payment authorization. That is expected for this service, but it grants account/payment authority.
Auth: `X-API-Key` or x402 payment (USDC on Base).
Use a dedicated Seisoai key with least privilege, quotas, and rotation; do not share broad account credentials with the agent.
Using agent-scoped routes could send prompts or task details to a specific remote agent workflow if explicitly requested.
The skill can call Seisoai agent-scoped orchestration routes, but it also documents restrictive checks such as exact agent binding, allowlists, and deny-by-default behavior.
`POST /api/gateway/agent/{agentId}/orchestrate` ... "Default posture: deny by default"Only use agent-scoped endpoints when the user names the exact agent or workflow, and verify the agent ID and allowed tools before invoking.