Back to skill

Security audit

Chords Fetcher

Security checks for vulnerabilities and agentic risk

Overview

The skill’s chord lookup behavior is mostly on-purpose, but it needs review because it runs a local command with user-supplied song text and uses external network lookups without tight scoping.

Install only if you are comfortable with the agent running a local Python script and sending song titles or artists to DuckDuckGo and chord websites. The maintainer should tighten the invocation to pass the song query as a single literal argument, declare the required exec and network scope, prefer HTTPS, and validate redirect targets.

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 (2)

T09 · Insecure Skill Coding Practices

Error
Location
SKILL.md:14
Finding

Unquoted User Input in Documented Shell Command

Content
View full analysis
``` The same unsafe invocation pattern is repeated in `README.md`: ```bash uv run python chords.py ``` ### Technical Analysis The skill instructs an AI agent to execute a shell command by substituting the user-provided song name and artist into an unquoted command template. If the agent constructs a command string from this template and sends it to a shell, shell metacharacters contained in the song query can be interpreted as shell syntax instead of ordinary Python arguments. Potentially dangerous syntax includes command separators, command substitution, pipes, redirections, and environment-variable expansion. The Python script itself does not invoke a shell, but the documented execution boundary is unsafe because an agent following the instructions may interpolate untrusted input directly into the command. Exploitability depends on whether the agent's `exec` implementation invokes a shell and whether it performs safe argument handling. The documentation neither requires an argument-array API nor mandates shell escaping. ### Attack Path 1. An attacker asks the agent to retrieve chords for a song whose purported name contains shell metacharacters and an additional command. 2. The agent follows `SKILL.md` and substitutes the complete attacker-controlled query into the documented command. 3. The agent submits the constructed command string through a shell-backed execution tool. 4. The shell interprets the injected metacharacters. 5. The additional command executes under the same operating-system identity and permissions as the AI agent. ### Impact Assessment Successful exploitation can execute arbitrary local commands with the privileges ...[truncated 450 chars]
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
chords.py:22
Finding

Destination Allowlist Can Be Bypassed Through HTTP Redirects

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (5)

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
91% confidence
Finding

The skill instructs use of an exec-run Python script that performs network access, but the manifest declares no permissions or allowed-tools scope. That mismatch weakens policy enforcement and reviewability, making it easier for a skill to exfiltrate user queries or fetch untrusted remote content without an explicit declaration. In this context the network use is expected for chord lookup, but undeclared capability still increases risk because user song requests are sent to external services.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
86% confidence
Finding

The invocation rule is very broad: any request for song chords triggers execution of the script. Broad trigger conditions can cause the agent to over-invoke external/networked tooling for casual or ambiguous requests, unnecessarily sending user input to third-party sites and expanding attack surface through frequent execution of code and scraping logic.

Content

No source excerpt is available for this finding.

Ae4

Medium
Category
analysis-evasion
Confidence
80% confidence
Finding

Suspicious Unicode normalization or mixed-script content

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The request hard-codes Accept-Language: ru-RU,ru..., forcing a Russian locale for all fetches without user consent or configuration. While this is not code-execution dangerous, it can cause privacy leakage about assumed locale, unexpected content selection, and reduced transparency for users in other regions.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
93% confidence
Finding

The skill description does not warn users that their song query will be sent to DuckDuckGo and third-party chord websites. This is primarily a transparency and privacy issue: users may unknowingly disclose listening interests or typed query content to external services, which is more concerning because the skill is explicitly network-enabled.

Content

No source excerpt is available for this finding.