Back to skill

Security audit

Validator Tool

Security checks for vulnerabilities and agentic risk

Overview

The skill is not malicious, but it needs review because its documentation overstates supported validation features and the code includes undocumented credit-card validation that can expose entered numbers in logs or process arguments.

Review before installing. Treat this as a small local validator script, not as the full-featured validation tool described in the skill text. Do not pass real payment-card data or other secrets to it unless the code is changed to accept safer input and redact output, and do not rely on the documented path, schema, batch, strict, or JSON-output behavior until those features are actually implemented.

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 (1)

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/validate.py:128
Finding

Plaintext Exposure of Credit-Card Data Through Process Arguments and Console Output

Content
View full analysis
None: """Try all validators.""" validators = [ ("Email", validate_email), ("Phone", validate_phone), ("URL", validate_url), ("IP", validate_ip), ("Credit Card", validate_credit_card), ("JSON", validate_json), ("UUID", validate_uuid), ("Hex", validate_hex), ] print(f"Testing '{value}' with all validators:") print("-" * 50) ``` The sensitive value is obtained directly from a command-line argument: ```python validator_type = sys.argv[1].lower() value = sys.argv[2] ``` ### Technical Analysis The tool supports credit-card validation even though this capability is not declared in `SKILL.md`. Credit-card numbers are supplied through command-line arguments and may therefore be exposed through shell history, process inspection facilities, terminal recordings, or execution telemetry. The `validate_all` function creates an additional disclosure by printing the complete supplied value without masking it: ```python print(f"Testing '{value}' with all validators:") ``` If the supplied value is a payment-card number, the full number is written to standard output. In CI systems, automation frameworks, agent transcripts, or applications that collect subprocess output, this value may be retained in logs or artifacts. The implementation does not redact, tokenize, or otherwise minimize sensitive data. ### Attack Path 1. A user or automated process invokes the tool with a card number, for example: ```bash python3 scripts/validate.py all 4532015112830366 ``` 2. The complete card number is placed in the process argument vector and may also be recorded in shell history. 3. A local user or monitoring process with sufficient process-observat ...[truncated 1205 chars]
Remediation
View remediation
str: digits = re.sub(r"[\s-]", "", value) return "*" * max(0, len(digits) - 4) + digits[-4:] ``` 5. Change `validate_all` so it identifies validators without reproducing the original input: ```python print("Testing supplied value with all validators:") ``` 6. Ensure verbose, JSON, error, and exception output paths apply the same redaction policy. 7. Configure CI and application logging to avoid retaining sensitive command arguments and subprocess output. 8. Add automated tests confirming that complete card numbers never appear in stdout, stderr, structured output, or exception messages. ]]>
Vulnerability Patterns
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The skill documentation advertises specific validation capabilities and batch processing semantics that apparently do not match actual behavior. This can cause downstream agents or users to rely on missing checks such as file path validation, or assume broader validation guarantees than are truly provided, leading to security control gaps and unsafe input handling in automation pipelines.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The manifest says the skill validates JSON, email, URL, file paths, IP addresses, and phone numbers. This file also implements credit card, UUID, hexadecimal, and date validation, which expands the skill's behavior beyond the described set, while file path validation is notably absent.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.