Agent Bazaar

Security checks across static analysis, malware telemetry, and agentic risk

Overview

This is a clear Agent Bazaar guide, but it may let an agent spend USDC from a connected wallet on paid API calls without explicit spending limits or per-call approval.

Use this only with clear payment controls: start in demo mode, require approval before real USDC payments, set a budget for each task, and avoid sending secrets or confidential data to Agent Bazaar endpoints.

Static analysis

No static analysis findings were reported for this release.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Risk analysis

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.

What this means

If a wallet is connected, the agent could initiate paid Agent Bazaar calls and spend wallet funds during a task.

Why it was flagged

This directs the agent to use an existing wallet context to execute USDC payments. The provided artifacts do not clearly require per-transaction user confirmation, wallet spending caps, or workflow-level budgets.

Skill content
Wallet already configured → Use the existing lobster.cash wallet... lobster.cash executes the payment and returns proof
Recommendation

Require explicit user approval before each real payment, set a maximum budget per task/workflow, prefer demo mode first, and declare the wallet/payment dependency clearly in metadata.

What this means

A complex request could trigger several paid API calls, causing cumulative charges beyond what the user expected.

Why it was flagged

The skill encourages multi-step paid tool chains. Without explicit approval, budget, or stop conditions in the provided artifacts, task expansion could lead to repeated paid calls.

Skill content
Chain skills for complex tasks. Each call is a separate payment via lobster.cash
Recommendation

Add clear workflow controls: show the total estimated cost, ask for approval before paid calls, cap retries/chains, and stop when the budget is reached.

What this means

Selected task data may be transmitted to Agent Bazaar endpoints for processing.

Why it was flagged

The documented endpoints send user-provided material such as source code, text, URLs, prompts, and portfolio details to Agent Bazaar. This is expected for the marketplace integration, but it is still an external data flow users should notice.

Skill content
Base URL: `https://agent-bazaar.com` ... `code` (required): Source code to review
Recommendation

Avoid submitting secrets, private keys, proprietary code, confidential documents, or sensitive financial details unless the user has approved that disclosure.