Back to skill

Security audit

MovieFetch

Security checks for vulnerabilities and agentic risk

Overview

MovieFetch mostly does what it says, but its remove action can delete media files by default from an ambiguous title match without a clear confirmation step.

Install only if you are comfortable giving this skill API access to your media stack. Treat remove/cancel/delete prompts as dangerous until the skill is changed to require exact movie identity and explicit confirmation, and use HTTPS or loopback-only service URLs for credentials.

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 (3)

T09 · Insecure Skill Coding Practices

Error
Location
scripts/remove_movie.py:18
Finding

Ambiguous title matching can delete unintended movie files

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/check_download.py:12
Finding

Service credentials may be transmitted over plaintext HTTP

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Note
Location
scripts/check_ott.py:7
Finding

TMDB API key is exposed in URL query parameters

Content
View full analysis
`. 3. A local debugger, instrumented HTTP client, trusted proxy, monitoring pro ...[truncated 632 chars]
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Taint TrackingDirect Taint Flow, Variable-Mediated Taint Flow, Credential Exfiltration Chain
Findings (33)

Tainted flow: 'params' from os.environ.get (line 22, credential/environment) → requests.get (network output)

Critical
Category
Data Flow
Confidence
90% confidence
Finding

Credentials or environment variables flow to a network sink. This is a high-confidence indicator of credential exfiltration.

Content

Scanner excerpt · scripts/check_ott.py (reported line 14)May include surrounding context.

python
"""Search for a movie or TV series on TMDB."""
    url = f"{TMDB_URL}/search/{media_type}"
    params = {"api_key": TMDB_API_KEY, "query": query}
    resp = requests.get(url, params=params, timeout=10)
    resp.raise_for_status()
    results = resp.json().get("results", [])
    return results[0] if results else None

Tainted flow: 'params' from os.environ.get (line 22, credential/environment) → requests.get (network output)

Critical
Category
Data Flow
Confidence
90% confidence
Finding

Credentials or environment variables flow to a network sink. This is a high-confidence indicator of credential exfiltration.

Content

Scanner excerpt · scripts/check_ott.py (reported line 23)May include surrounding context.

python
"""Search for a movie or TV series on TMDB."""
    url = f"{TMDB_URL}/search/{media_type}"
    params = {"api_key": TMDB_API_KEY, "query": query}
    resp = requests.get(url, params=params, timeout=10)
    resp.raise_for_status()
    results = resp.json().get("results", [])
    return results[0] if results else None

Tainted flow: 'url' from os.environ.get (line 30, credential/environment) → requests.get (network output)

Critical
Category
Data Flow
Confidence
90% confidence
Finding

Credentials or environment variables flow to a network sink. This is a high-confidence indicator of credential exfiltration.

Content

Scanner excerpt · scripts/check_plex.py (reported line 20)May include surrounding context.

python
"""Returns the section key for the configured movie library."""
    url = f"{PLEX_URL}/library/sections"
    headers = {"X-Plex-Token": PLEX_TOKEN, "Accept": "application/json"}
    resp = requests.get(url, headers=headers, timeout=5)
    resp.raise_for_status()
    for section in resp.json()["MediaContainer"]["Directory"]:
        if section["title"].lower() == PLEX_LIBRARY.lower():

Tainted flow: 'url' from os.environ.get (line 30, credential/environment) → requests.get (network output)

Critical
Category
Data Flow
Confidence
90% confidence
Finding

Credentials or environment variables flow to a network sink. This is a high-confidence indicator of credential exfiltration.

Content

Scanner excerpt · scripts/check_plex.py (reported line 33)May include surrounding context.

python
url = f"{PLEX_URL}/library/sections/{library_key}/search"
    headers = {"X-Plex-Token": PLEX_TOKEN, "Accept": "application/json"}
    params = {"query": title, "type": 1}  # type 1 = movies
    resp = requests.get(url, headers=headers, params=params, timeout=5)
    resp.raise_for_status()
    results = resp.json()["MediaContainer"].get("Metadata", [])
    return results

Tainted flow: 'HEADERS' from os.environ.get (line 15, credential/environment) → requests.get (network output)

Critical
Category
Data Flow
Confidence
90% confidence
Finding

Credentials or environment variables flow to a network sink. This is a high-confidence indicator of credential exfiltration.

Content

Scanner excerpt · scripts/remove_movie.py (reported line 20)May include surrounding context.

python
def find_movie_in_library(title: str):
    """Search the Radarr library for a movie by title (case-insensitive partial match)."""
    resp = requests.get(f"{RADARR_URL}/movie", headers=HEADERS, timeout=10)
    resp.raise_for_status()
    title_lower = title.lower()
    matches = [m for m in resp.json() if title_lower in m["title"].lower()]

Tainted flow: 'HEADERS' from os.environ.get (line 18, credential/environment) → requests.get (network output)

Critical
Category
Data Flow
Confidence
90% confidence
Finding

Credentials or environment variables flow to a network sink. This is a high-confidence indicator of credential exfiltration.

Content

Scanner excerpt · scripts/request_movie.py (reported line 23)May include surrounding context.

python
def lookup_movie(title: str):
    """Search TMDB via Radarr — returns top match or None."""
    resp = requests.get(
        f"{RADARR_URL}/movie/lookup",
        headers=HEADERS,
        params={"term": title},

Tainted flow: 'HEADERS' from os.environ.get (line 18, credential/environment) → requests.get (network output)

Critical
Category
Data Flow
Confidence
90% confidence
Finding

Credentials or environment variables flow to a network sink. This is a high-confidence indicator of credential exfiltration.

Content

Scanner excerpt · scripts/request_movie.py (reported line 47)May include surrounding context.

python
def get_movie_id(tmdb_id: int):
    """Fetch internal Radarr movie ID from TMDB ID."""
    resp = requests.get(
        f"{RADARR_URL}/movie",
        headers=HEADERS,
        params={"tmdbId": tmdb_id},

Tainted flow: 'HEADERS' from os.environ.get (line 18, credential/environment) → requests.get (network output)

Critical
Category
Data Flow
Confidence
90% confidence
Finding

Credentials or environment variables flow to a network sink. This is a high-confidence indicator of credential exfiltration.

Content

Scanner excerpt · scripts/request_movie.py (reported line 61)May include surrounding context.

python
def search_and_download_release(movie_id: int, extra_words: set):
    """Search releases for a movie and queue the first one that matches all extra keywords."""
    resp = requests.get(
        f"{RADARR_URL}/release",
        headers=HEADERS,
        params={"movieId": movie_id},

Tainted flow: 'RADARR_URL' from os.environ.get (line 13, credential/environment) → requests.post (network output)

Critical
Category
Data Flow
Confidence
90% confidence
Finding

The destination URL is taken from RADARR_URL without validation, and the request includes an API key in headers. If an attacker can influence environment configuration or deployment settings, they can redirect this POST to an attacker-controlled host and capture the credential while also triggering unintended outbound requests.

Content

Scanner excerpt · scripts/request_movie.py (reported line 97)May include surrounding context.

python
"indexerId": best_release["indexerId"],
        "movieId": movie_id
    }
    dl_resp = requests.post(
        f"{RADARR_URL}/release",
        headers={**HEADERS, "Content-Type": "application/json"},
        json=dl_payload,

Tainted flow: 'payload' from os.environ.get (line 114, credential/environment) → requests.post (network output)

Critical
Category
Data Flow
Confidence
90% confidence
Finding

Credentials or environment variables flow to a network sink. This is a high-confidence indicator of credential exfiltration.

Content

Scanner excerpt · scripts/request_movie.py (reported line 128)May include surrounding context.

python
"searchForMovie": not is_specific_version
        },
    }
    resp = requests.post(
        f"{RADARR_URL}/movie",
        headers={**HEADERS, "Content-Type": "application/json"},
        json=payload,

Credential Access

High
Category
Privilege Escalation
Confidence
70% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · README.md (reported line 98)May include surrounding context.

md
- Radarr with API access
- qBittorrent Web UI enabled
- Plex server access token

### Environment Variables

Tp4

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

This variant of the mismatch again shows narrower implemented behavior than the manifest promises. Repeated discrepancies are a red flag for poor change control and make the skill context more dangerous because users expect coordinated Plex/Radarr/qBittorrent operations when only a subset may execute.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

This variant of the mismatch again shows narrower implemented behavior than the manifest promises. Repeated discrepancies are a red flag for poor change control and make the skill context more dangerous because users expect coordinated Plex/Radarr/qBittorrent operations when only a subset may execute.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
92% confidence
Finding

This variant of the mismatch again shows narrower implemented behavior than the manifest promises. Repeated discrepancies are a red flag for poor change control and make the skill context more dangerous because users expect coordinated Plex/Radarr/qBittorrent operations when only a subset may execute.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

This variant of the mismatch again shows narrower implemented behavior than the manifest promises. Repeated discrepancies are a red flag for poor change control and make the skill context more dangerous because users expect coordinated Plex/Radarr/qBittorrent operations when only a subset may execute.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding

This variant of the mismatch again shows narrower implemented behavior than the manifest promises. Repeated discrepancies are a red flag for poor change control and make the skill context more dangerous because users expect coordinated Plex/Radarr/qBittorrent operations when only a subset may execute.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

This variant of the mismatch again shows narrower implemented behavior than the manifest promises. Repeated discrepancies are a red flag for poor change control and make the skill context more dangerous because users expect coordinated Plex/Radarr/qBittorrent operations when only a subset may execute.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
89% confidence
Finding

This variant of the mismatch again shows narrower implemented behavior than the manifest promises. Repeated discrepancies are a red flag for poor change control and make the skill context more dangerous because users expect coordinated Plex/Radarr/qBittorrent operations when only a subset may execute.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
95% confidence
Finding

The script defaults delete_files to true, so a removal request will delete the movie record and its downloaded files unless the caller explicitly opts out. In a chat-triggered automation context, ambiguous user phrasing, LLM misunderstanding, or accidental invocation could cause irreversible data loss without confirmation.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The README explicitly advertises a remove capability that can delete files, but it does not clearly warn users about irreversible data loss or emphasize confirmation requirements. In a chat-driven skill, ambiguous natural-language commands increase the chance of accidental destructive actions, making this more dangerous than in a traditional UI with explicit safeguards.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
88% confidence
Finding

The skill advertises networked automation across Plex, Radarr, and qBittorrent but does not declare any explicit tool scope such as allowed tools or permissions. In a chat-triggered skill that can query services and potentially remove media, missing scope boundaries increases the risk of unintended environment access or overbroad network use if the runtime grants default capabilities.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The skill describes cleanup that removes movies from the Radarr library and may delete files, but it does not include an explicit warning or confirmation requirement. In a chat-controlled skill, destructive actions without prominent caution are dangerous because a vague or mistaken request could trigger irreversible media deletion.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

For a skill described as managing movie downloads and status/removal across Plex, Radarr, and qBittorrent, querying a separate TMDB service for OTT provider data is not directly necessary based on the manifest alone. The use of external API access and secret-backed lookups introduces a capability beyond the declared movie-management purpose.

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/check_ott.py (reported line 8)May include surrounding context.

python
import requests

TMDB_API_KEY = os.environ.get("TMDB_API_KEY", "")
TMDB_URL = "https://api.themoviedb.org/3"

def search_tmdb(query: str, media_type: str = "movie"):
    """Search for a movie or TV series on TMDB."""

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The manifest describes a chat skill for downloading, tracking, and removing movies in Plex, Radarr, and qBittorrent. This script instead searches TMDB and retrieves watch-provider/streaming availability data, which is a distinct end-user capability not mentioned in the skill description and not an obvious implementation detail of download/status/remove operations.

Content

No source excerpt is available for this finding.