ClawHub
Back to skill

Security audit

finance-markets-research

Security checks across malware telemetry and agentic risk

Overview

The skill is mostly a disclosed finance-data API helper, but its executable client can call arbitrary Crawlora API paths beyond the finance scope it advertises.

Review before installing if you need strict finance-only behavior. Use the script only for the documented finance, SEC, Congress, crypto, and PitchBook endpoints, keep the Crawlora key in secret storage or an environment variable, and avoid sending sensitive private prompts or identifiers to the third-party API.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (4)

Description-Behavior Mismatch

Medium
Confidence
91% confidence
Finding
The SEC `company_intelligence` endpoint supports `enrich=hiring` with ATS-specific parameters (`ats`, `careers_slug`, `tenant`, `datacenter`, `site`), which expands the skill beyond the stated finance/markets research scope into employment-intelligence collection. That scope creep increases privacy and data-governance risk because user prompts intended for market research could trigger collection of workforce or recruiting-related data not clearly disclosed by the manifest.

Description-Behavior Mismatch

Medium
Confidence
88% confidence
Finding
The helper accepts an arbitrary API path and examples for non-finance endpoints, so the skill can reach Crawlora capabilities beyond its declared finance-markets-research scope. In an agent environment, this scope mismatch weakens least-privilege assumptions and could let downstream prompts or wrappers use unrelated data sources through the same authenticated client.

Intent-Code Divergence

Medium
Confidence
85% confidence
Finding
The inline documentation presents this as a general-purpose Crawlora client rather than a finance-specific helper, contradicting the skill's stated purpose. That increases the chance that integrators or agent logic will invoke non-finance endpoints, effectively expanding the operational scope without any explicit guardrail in code.

Missing User Warnings

Low
Confidence
76% confidence
Finding
The reference states that requests are sent to `api.crawlora.net` with an API key header but provides no warning about external transmission, credential handling, logging, or data-sensitivity expectations. In a tool used by agents, missing disclosure and handling guidance can lead to operators sending sensitive user queries or identifiers to a third-party service without informed controls.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

Detected: suspicious.exposed_secret_literal

File appears to expose a hardcoded API secret or token.

Critical
Code
suspicious.exposed_secret_literal
Location
SKILL.md:23