Back to skill

Security audit

Cheapcharts

Security checks for vulnerabilities and agentic risk

Overview

This skill is a coherent CheapCharts deal lookup helper that makes disclosed public web requests and does not show hidden persistence, credential access, data theft, or destructive behavior.

Install this only if you are comfortable with the agent making public web requests to CheapCharts and, for optional compatibility or savings advice, IMDb or CheapCharts web pages. Treat prices and availability as informational and verify them before buying; configure any recurring monitoring yourself rather than letting an agent silently add schedules.

Vulnerability Patterns
  • 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
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Findings (13)

External Script Fetching

High
Category
Supply Chain
Confidence
90% confidence
Finding

Remote code is downloaded and executed. This bypasses code review and could introduce malicious code.

Content

Scanner excerpt · RECIPES.md (reported line 23)May include surrounding context.

One call per endpoint (quick smoke tests)

bash
curl -s "https://buster.cheapcharts.de/v1/gptapi/Search.php?action=search&store=itunes&country=us&itemType=all&query=Fight%20Club&limit=5"
bash

External Script Fetching

High
Category
Supply Chain
Confidence
90% confidence
Finding

Remote code is downloaded and executed. This bypasses code review and could introduce malicious code.

Content

Scanner excerpt · RECIPES.md (reported line 87)May include surrounding context.

bash
# Step 1 - find IMDb ID
imdb=$(curl -s "https://buster.cheapcharts.de/v1/gptapi/Search.php?action=search&store=itunes&country=us&itemType=all&query=The%20Dark%20Knight&limit=1" \
  | python -c "import sys,json; print(json.load(sys.stdin)['results'][0].get('imdbId',''))")
echo "IMDb: $imdb"

External Script Fetching

High
Category
Supply Chain
Confidence
90% confidence
Finding

Remote code is downloaded and executed. This bypasses code review and could introduce malicious code.

Content

Scanner excerpt · RECIPES.md (reported line 94)May include surrounding context.

md
# Step 2 - query each store in parallel (only iTunes reliably supports prices; others via Topseller)
for store in itunes amazon vudu googlePlay; do
  echo "=== $store ==="
  curl -s "https://buster.cheapcharts.de/v1/gptapi/Prices.php?action=getPrices&store=$store&country=us&itemType=buymovies&imdbIDs=$imdb" \
    | python -c "
import sys, json
r = json.load(sys.stdin)

External Script Fetching

High
Category
Supply Chain
Confidence
90% confidence
Finding

Remote code is downloaded and executed. This bypasses code review and could introduce malicious code.

Content

Scanner excerpt · RECIPES.md (reported line 111)May include surrounding context.

"Complete TV series on sale" (filter to bundle deals, avoid per-season noise)

bash
curl -s "https://buster.cheapcharts.de/v1/gptapi/Deals.php?action=getDeals&store=itunes&country=us&itemType=seasons&genre=All&sort=greatestSavings&limit=50" | python -c "
import sys, json
items = json.load(sys.stdin)['results']['seasons']
bundles = [i for i in items if i.get('isBundle') == 1]

External Script Fetching

High
Category
Supply Chain
Confidence
90% confidence
Finding

Remote code is downloaded and executed. This bypasses code review and could introduce malicious code.

Content

Scanner excerpt · RECIPES.md (reported line 126)May include surrounding context.

"Cross-store top sellers today"

bash
curl -s "https://buster.cheapcharts.de/v1/gptapi/Topseller.php?action=getTopsellerForStartpage&country=us&store=itunes,amazon,vudu,googlePlay&maxItemCount=5" | python -c "
import sys, json
r = json.load(sys.stdin)
if r.get('status') != 'success': sys.exit(r.get('message'))

External Script Fetching

High
Category
Supply Chain
Confidence
90% confidence
Finding

Remote code is downloaded and executed. This bypasses code review and could introduce malicious code.

Content

Scanner excerpt · RECIPES.md (reported line 145)May include surrounding context.

bash
# Step 1 - Search to get IMDb ID
curl -s "https://buster.cheapcharts.de/v1/gptapi/Search.php?action=search&store=itunes&country=us&itemType=all&query=The%20Dark%20Knight&limit=3"
bash

External Script Fetching

High
Category
Supply Chain
Confidence
90% confidence
Finding

Remote code is downloaded and executed. This bypasses code review and could introduce malicious code.

Content

Scanner excerpt · RECIPES.md (reported line 211)May include surrounding context.

"Latest new-release movies on sale"

bash
curl -s "https://buster.cheapcharts.de/v1/gptapi/Deals.php?action=getDeals&store=itunes&country=us&itemType=buymovies&genre=All&sort=releaseDate&limit=500" | python -c "
import sys, json
items = json.load(sys.stdin)['results']['buymovies']
real = [i for i in items if not i.get('releaseDate','').startswith('2030')]

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The manifest advertises multi-store price lookups, but the recipe itself states only iTunes reliably supports direct prices while other stores require a different endpoint or may fail. This mismatch can cause the agent to make unsupported cross-store requests and return misleading or inconsistent results to users.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The skill includes scraping IMDb with curl and grep for a Movies Anywhere studio lookup, which is outside the stated CheapCharts API deal-lookup purpose. Adding unrelated external web scraping broadens network access, introduces dependence on a third-party site’s HTML, and creates a path for off-scope data retrieval and brittle behavior.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The recipes explicitly direct use of an unofficial internal endpoint, DetailData.php, discovered via website network calls and outside the documented GPT API surface. Depending on an undocumented endpoint expands the skill’s effective capability beyond the declared interface, creates supply-chain and stability risk, and may expose fields or behaviors not intended for agent use.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
91% confidence
Finding

The skill clearly instructs use of a public external API and bundled scripts that perform network requests, but it does not declare any explicit tool scope such as allowed network access. That mismatch is a real security and governance issue because an agent runtime may grant broader capabilities than reviewers or policy expect, making external data exfiltration or unintended outbound requests harder to constrain and audit.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The activation guidance says to load this file when a conversation turns to broad adjacent topics like maximizing savings, cross-store ownership, seasonal timing, or video games. Those triggers are vague enough that the extras content may be pulled into contexts where it is unnecessary, increasing the attack surface and the chance the agent follows lower-priority or out-of-scope guidance. In this skill, the file also contains non-core fallback and adjacent-topic instructions, so unintended loading can influence behavior beyond the core movie/TV deal lookup flow.

Content

No source excerpt is available for this finding.

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
75% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · references/PITFALLS.md (reported line 72)May include surrounding context.

md
### 15. Always check response.status before iterating results

Every endpoint returns `{"status":"success", ...}` on success or `{"status":"error", "message":"..."}` on failure. If you iterate `results` without checking status, a 400 error with empty `results: {}` silently yields no data - which looks identical to "no deals found." This is how the 2026-06-21 "no drops today" bug happened: passing `itemType=buymovies` to DetailData returned a status=error, the script iterated an empty dict, and reported "nothing today" when The Harvest had actually just dropped. **Pattern: `if response.get('status') != 'success': print(response.get('message')); return`.**

### 16. has4K=1 filter does NOT work on Deals/Charts