Token Shark

Security checks across malware telemetry and agentic risk

Overview

The skill does not show data theft or trading code, but it appears to present mock token data as real-time investment analysis, which could mislead users.

Only install this as a demo or toy tool unless the author clearly documents and implements real data sources. Do not rely on its prices, risk scores, or investment suggestions for trading, and be cautious about sending crypto sponsorship payments for claimed unlimited queries.

VirusTotal

63/63 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.

#
ASI09: Human-Agent Trust Exploitation
Medium
What this means

A user could mistake hardcoded/demo token prices, liquidity, and risk scores for real market intelligence and make poor financial decisions.

Why it was flagged

The code itself says the token feed is simulated and should actually come from a DEX API, while the skill advertises real-time monitoring, risk scoring, and market analysis.

Skill content
// 模拟新代币数据 (实际应从 DEX API 获取)
Recommendation

Clearly label all outputs as mock/demo data unless live DEX, on-chain, and contract-analysis sources are actually implemented and disclosed.

#
ASI09: Human-Agent Trust Exploitation
Medium
What this means

The skill may produce confident-looking risk ratings and investment suggestions that are not grounded in actual token safety data.

Why it was flagged

Risk assessment is based on the last character of the provided address rather than real contract, liquidity, holder, team, or community analysis.

Skill content
const lastChar = address.slice(-1); ... const isLowRisk = ['0', '1', '2', 'a', 'b', 'c'].includes(lastChar);
Recommendation

Do not present these scores as real risk assessments; either implement verifiable data sources or restrict the skill to demonstration/testing use.

#
ASI06: Memory and Context Poisoning
Low
What this means

Someone with local file access could see which token alerts the user created.

Why it was flagged

The alert feature persists token addresses and target prices to a local JSON file.

Skill content
const ALERTS_FILE = path.join(DATA_DIR, 'alerts.json'); ... fs.writeFileSync(ALERTS_FILE, JSON.stringify(data, null, 2));
Recommendation

Treat alert data as local watchlist information and delete data/alerts.json if you no longer want it retained.