Claw Credit by t54

Security checks across malware telemetry and agentic risk

Overview

This credit-payment skill is coherent, but it asks the agent to use an external SDK that can spend credit and collect sensitive local agent prompts, traces, and context with unclear limits.

Before installing, confirm the exact SDK version and source, understand the credit and repayment terms, require approval for every paid call, set strict spending limits, and do not allow broad transcript or prompt-directory collection unless you have reviewed and redacted the data that will be shared.

VirusTotal

66/66 vendors flagged this skill as clean.

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.

#
ASI03: Identity and Privilege Abuse
High
What this means

An agent could incur credit-backed charges for third-party services, creating a repayment obligation for the user.

Why it was flagged

The skill delegates payment and credit-spending authority to the SDK/agent, but the artifacts do not show clear per-transaction user approval, merchant limits, spending caps, or repayment controls.

Skill content
The ability to call any x402-enabled API or service without upfront payment ... Repay later ... You can now call credit.pay() directly
Recommendation

Use only with explicit spending limits, merchant allowlists, clear repayment terms, and user confirmation before each paid call.

#
ASI06: Memory and Context Poisoning
High
What this means

Private prompts, agent instructions, session history, workspace content, or environment details could be exposed during registration or credit evaluation.

Why it was flagged

The skill instructs the SDK to gather system prompts, traces, environment details, session transcripts, and prompt directories, but does not define exclusions, redaction, retention, or exactly what is sent to the credit service.

Skill content
Enable tracing to capture reasoning and execution context ... The SDK auto-collects prompt and environment details from the trace ... transcriptDirs: ["/path/to/.openclaw/agents/main/sessions"], promptDirs: ["/path/to/openclaw/workspace", "/path/to/.openclaw/agents/main/agent"]
Recommendation

Review and restrict the configured directories, confirm what the SDK uploads, redact secrets, and require clear retention/deletion terms before registering.

#
ASI04: Agentic Supply Chain Vulnerabilities
Medium
What this means

The most sensitive behavior depends on code that was not available for this review and could change over time if installed without pinning.

Why it was flagged

The reviewed package contains only instructions and relies on an external, unpinned npm SDK for the core behavior. That SDK is not included in the artifacts, yet it is expected to handle payment authority, saved tokens, local traces, and prompt/context collection.

Skill content
npm install @t54-labs/clawcredit-sdk
Recommendation

Pin and review the SDK version, verify its publisher and source, and avoid granting payment or local-context access until the installed package is audited.