Back to skill

Security audit

india-market-intelligence

Security checks for vulnerabilities and agentic risk

Overview

This skill appears to fetch market data and news as advertised, but it presents simple heuristics as AI-backed trading guidance and uses unpinned runtime dependencies.

Install only if you are comfortable with a user-run Python tool that contacts third-party market and news services and auto-resolves dependencies. Treat all trading signals, probabilities, risk scores, position-sizing guidance, and BUY/SELL/LONG/SHORT labels as unvalidated heuristic commentary, not financial advice, and verify market data independently before acting.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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)

T08 · Insecure Dependencies

Warning
Location
scripts/market_analysis.py:3
Finding

Unpinned Dependencies Are Automatically Resolved and Installed at Runtime

Content
View full analysis
Remediation
View remediation
", # "pandas==", # "rich==", # "feedparser==", # "numpy==" # ] ``` 2. **Generate and commit a lockfile** that records the complete transitive dependency graph. 3. **Require package integrity verification** using cryptographic hashes where the selected installation workflow supports it. 4. **Remove unused dependencies and imports**: - Remove `requests` from `scripts/market_analysis.py` unless direct HTTP requests are added. - Remove `beautifulsoup4` from the main script's dependency declaration unless HTML parsing is required. - Remove unused `requests` and `BeautifulSoup` imports from `scripts/news_fetcher.py` if RSS parsing remains its only retrieval mechanism. 5. **Use a dedicated virtual environment or sandbox** with no administrative privileges and minimal access to credentials and sensitive files. 6. **Restrict package sources** to an explicitly configured trusted package index. Do not permit fallback to arbitrary or unreviewed indexes. 7. **Add automated dependency controls**, including vulnerability scanning, update review, provenance verification, and periodic lockfile regeneration. 8. **Review dependency updates before deployment** rather than automatically consuming the latest available release during each invocation. ]]>
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • 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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (11)

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The skill description promises real-time market data, AI analysis, probability predictions, risk scoring, and trading decisions, while the detected behavior reportedly only consumes RSS news feeds and lacks the claimed analytics. This mismatch is dangerous because downstream users or agents may trust outputs as quantitatively grounded financial intelligence when they are not, leading to unsafe decision-making and concealment of the skill's actual behavior.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The manifest describes an India-focused market intelligence tool centered on Indian markets, though it may include some global context. This Quickstart presents the skill as 'Global Market Intelligence' and promotes commands and symbols for broad global indices, commodities, and even non-Indian tickers like AAPL, which materially expands the apparent scope beyond the manifest's stated focus.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
86% confidence
Finding

The README explicitly advertises AI-driven probability predictions, risk scoring, trading recommendations, position sizing, and stop-loss calculations, while only providing a broad 'not financial advice' disclaimer. That framing can cause users to over-rely on opaque model outputs for real financial decisions without clear warnings about uncertainty, model error, or the need for independent verification.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
91% confidence
Finding

The skill advertises network-backed functionality but does not declare any tool scope such as permissions or allowed-tools. This creates an authorization and transparency gap: an agent or reviewer cannot clearly determine what external access the skill is intended to use, increasing the chance of unintended network access or policy bypass in environments that rely on manifest declarations.

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 this skill is India-focused, with global indices as supporting context for Indian stock market analysis. This file instead presents the skill as a comprehensive global market intelligence product, emphasizing US, European, Asian, commodities, currencies, and broad global use cases as core functionality rather than India-centered analysis.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The skill metadata promises AI-powered analysis and trading decisions, but this section generates deterministic, rule-based narrative output from simple thresholds. In a financial decision-support context, overstating analytical sophistication can mislead users into placing undue trust in outputs that lack validated modeling, uncertainty handling, or suitability checks.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

This code presents probability predictions and market direction assessments as if they are analytically robust, but the values are derived from hard-coded adjustments to RSI, momentum, and volatility. In a market-intelligence skill, fabricated precision and pseudo-probabilities can create false confidence and lead users to take risky actions based on unvalidated scoring logic.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

This section emits action-oriented labels such as BUY, SELL, LONG, SHORT, REDUCE, and EXPAND from simplistic thresholds, while the overall skill branding frames the output as comprehensive market intelligence. In the financial context, that mismatch materially increases the risk that users treat generic threshold logic as trustworthy trading advice, potentially causing financial harm.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The code generates explicit BUY/SELL/LONG/SHORT-style recommendations without verifying user intent, risk tolerance, investment horizon, jurisdictional constraints, or whether the user asked for prescriptive advice. In a trading skill, unsolicited prescriptive outputs are especially dangerous because they can drive unsuitable financial actions and increase legal/compliance exposure.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Low
Category
Not specified by scanner
Confidence
86% confidence
Finding

The file structure and surrounding documentation identify the implementation as 'global-market-intelligence-1.0.0' with corresponding scripts, while the manifest identifies the skill as 'india-market-intelligence'. This is more than incomplete documentation: it actively presents a different skill identity and scope than the manifest under review.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
78% confidence
Finding

This markdown file gives direct run commands for scripts that, by the document's own description, fetch market data and aggregate news from multiple external sources using requests, RSS, and web scraping. While the file includes financial disclaimers, it does not warn users that running the skill will make outbound network requests to third-party services and may expose IP, request metadata, or queried symbols/watchlists.

Content

No source excerpt is available for this finding.