Back to skill

Security audit

Meme Coin Scanner

Security checks for vulnerabilities and agentic risk

Overview

The advertised meme-coin scanner mostly matches its main script, but the package also includes an undisclosed unrelated security utility that can persist user inputs locally.

Review this package before installing. The main meme-token scanner uses DexScreener and appears aligned with the advertised purpose, but the extra script.sh utility is undisclosed, unrelated, and can store user-supplied arguments in a local history file; do not use its hash, encrypt, or password commands for sensitive values.

Vulnerability Patterns
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
Findings (2)

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/script.sh:16
Finding

Plaintext Persistence of Potentially Sensitive Command Arguments

Content
View full analysis

Vulnerability Details

File Location: scripts/script.sh, lines 16-17 and 51-65
Vulnerability Type: Plaintext sensitive-data logging
Risk Level: Medium

Vulnerable Code

bash
_log() { echo "$(date '+%m-%d %H:%M') $1: $2" >> "$DATA_DIR/history.log"; }
bash
cmd_encrypt() {
    echo "  Encrypting: $1"
    _log "encrypt" "${1:-}"
}

cmd_hash() {
    echo "$1" | sha256sum | cut -d" " -f1
    _log "hash" "${1:-}"
}

Technical Analysis

The _log function writes the first command argument directly to a persistent plaintext file. Both cmd_encrypt and cmd_hash pass their original input to this function without redaction.

Users may reasonably provide passwords, tokens, confidential strings, or other secrets to commands named encrypt and hash. Although cmd_hash prints a SHA-256 digest, the unmodified input is subsequently retained in history.log, defeating the confidentiality normally expected from hashing.

The data directory and history file are created without an explicit restrictive umask or file mode. Their actual permissions therefore depend on the invoking environment. The utility is also not declared in SKILL.md and is unrelated to the stated meme-coin scanning functionality, so persistent argument collection exceeds the minimum data access and storage required by the Skill.

Attack Path

  1. A user invokes scripts/script.sh encrypt or scripts/script.sh hash with a sensitive value.
  2. The command passes the original value to _log.
  3. _log appends the command name and plaintext value to $DATA_DIR/history.log.
  4. The value remains on disk after the command terminates.
  5. A local user, process, backup system, or later compromise with access to that file can recover the sensitive value.

Impact Assessment

This issue can disclose secrets supplied as command arguments. Its scope is primarily local confidentiality: it does not independently g ...[truncated 351 chars]

Remediation
View remediation

Remediation Suggestions

  1. Remove scripts/script.sh if it is not required for the declared meme-coin scanning functionality.

  2. Never log raw arguments supplied to encryption, hashing, authentication, or credential-related commands.

  3. If operational logging is necessary, retain only non-sensitive metadata, such as the command name, timestamp, and success status.

  4. Establish restrictive permissions before creating persistent data:

    bash
    umask 077
    mkdir -p "$DATA_DIR"
    touch "$DATA_DIR/history.log"
    chmod 600 "$DATA_DIR/history.log"
    
  5. Accept sensitive input through standard input or an interactive prompt rather than command-line arguments, which may also be exposed through shell history and process listings.

  6. Document all retained data, its purpose, location, permissions, and retention period.

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/script.sh:67
Finding

Security Passwords Generated with a Predictable Pseudorandom Generator

Content
View full analysis

Vulnerability Details

File Location: scripts/script.sh, lines 67-73
Vulnerability Type: Non-cryptographic random number generation
Risk Level: Medium

Vulnerable Code

bash
cmd_password() {
    python3 << 'PYEOF'
import random, string
print("".join(random.choices(string.ascii_letters + string.digits + "!@#", k=16)))
PYEOF
    _log "password" "${1:-}"
}

Technical Analysis

Python's random module uses a deterministic pseudorandom number generator intended for simulation and general-purpose sampling, not credential generation. It does not provide the unpredictability required for passwords, tokens, or other security secrets.

If an attacker can infer or recover the generator state, outputs produced by random.choices may be reproducible or predictable. Labeling this operation as a password generator encourages users to rely on output that lacks a cryptographically secure source of randomness.

This password-generation feature is not part of the meme-coin scanner functionality declared in SKILL.md, so it unnecessarily expands the Skill's security-sensitive behavior.

Attack Path

  1. A user invokes the bundled password command and trusts the output as a secure credential.
  2. The script generates the value using Python's non-cryptographic random module.
  3. The user assigns the generated value to an account or service.
  4. An attacker capable of recovering or sufficiently constraining the PRNG state generates candidate outputs.
  5. A matching candidate may be used to authenticate as the victim where no additional authentication controls prevent access.

Exploitation requires enough information or access to infer the generator state; the code alone does not provide remote access or automatic state recovery.

Impact Assessment

Successful prediction could expose any account protected by a generated password. Resulting privileges are limited to those granted by the affected ...[truncated 252 chars]

Remediation
View remediation

Remediation Suggestions

  1. Remove the password generator if it is not required by the declared Skill functionality.

  2. If retained, replace random with Python's cryptographically secure secrets module:

    python
    import secrets
    import string
    
    alphabet = string.ascii_letters + string.digits + "!@#"
    print("".join(secrets.choice(alphabet) for _ in range(20)))
    
  3. Use a sufficiently long output appropriate to the target service's requirements.

  4. Recommend unique credentials stored in a reputable password manager.

  5. Add automated checks that reject use of random, timestamp-derived seeds, or other deterministic sources for credential generation.

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (9)

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

A description-behavior mismatch is dangerous because it can hide capabilities unrelated to the stated meme-coin scanning purpose, such as generic system/security utilities, logging, hashing, password generation, or encryption helpers. When a skill's real behavior exceeds user expectations, it undermines trust and can enable stealthy misuse, privilege creep, or collection of local data under the cover of a benign token-analysis workflow.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

The implemented script does not perform meme-coin, rug-pull, liquidity-lock, or on-chain audit analysis as described in the skill metadata; instead it exposes a generic local security utility interface. This capability mismatch is dangerous because it can mislead users and orchestrators into invoking unrelated functionality under a trusted crypto-scanning label, increasing the chance of unintended local actions and improper trust decisions.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The script header and surrounding documentation present the tool as a meme-coin scanner while the code implements generic security helper operations. This deceptive or materially inaccurate labeling undermines user consent and safe tool selection, and in context makes the mismatch more dangerous because users expect blockchain-analysis behavior, not local utility and logging actions.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
88% confidence
Finding

The skill advertises shell commands and the analyzer detected network-capable behavior, but the manifest declares no explicit tool scope or permissions. That creates an authorization ambiguity where an agent or user may invoke network-reaching code without clear least-privilege boundaries, increasing the chance of unintended external access or data exposure.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · scripts/meme.sh (reported line 39)May include surrounding context.

sh
# Try DexScreener for basic info
try:
    url = "https://api.dexscreener.com/latest/dex/tokens/{}".format(token)
    req = Request(url)
    req.add_header("User-Agent", "MemeCoinScanner/1.0")
    resp = urlopen(req, timeout=15)

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · scripts/meme.sh (reported line 174)May include surrounding context.

sh
# Try DexScreener for basic info
try:
    url = "https://api.dexscreener.com/latest/dex/tokens/{}".format(token)
    req = Request(url)
    req.add_header("User-Agent", "MemeCoinScanner/1.0")
    resp = urlopen(req, timeout=15)

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · scripts/meme.sh (reported line 280)May include surrounding context.

sh
# Try DexScreener for basic info
try:
    url = "https://api.dexscreener.com/latest/dex/tokens/{}".format(token)
    req = Request(url)
    req.add_header("User-Agent", "MemeCoinScanner/1.0")
    resp = urlopen(req, timeout=15)

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The logging helper records command names and user-supplied arguments to a persistent history file without disclosure, and some commands may receive sensitive strings as inputs. This creates a privacy and security risk because secrets, token identifiers, investigative targets, or other sensitive operational data may be stored on disk unexpectedly and later exposed to other local users, backups, or forensic collection.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

Encryption, hashing, and password-generation functions are unrelated to a meme-coin scam scanner and materially expand the skill's capability surface beyond its stated purpose. In an agent ecosystem, unjustified extra capabilities are risky because they can be abused as covert utility functions, enable policy evasion, or cause users to trust a tool for actions they did not authorize.

Content

No source excerpt is available for this finding.