Back to skill

Security audit

Fzf

Security checks for vulnerabilities and agentic risk

Overview

This fzf skill is mostly a legitimate command-line guide, but it gives an AI broad permission to install tools, edit persistent shell startup files, and add risky command shortcuts without enough warning or confirmation.

Review before installing. Do not let an agent automatically edit your shell startup files or add the fh/fkill-style functions without seeing the exact diff. Prefer package-manager installation, avoid unverified releases/latest binary installs, and treat examples that kill processes, stop containers, drop/pop stashes, checkout files, or delete files as admin/destructive operations that need explicit confirmation.

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
guides/01-installation.md:86
Finding

Unverified Installation of a Mutable Remote Executable

Content
View full analysis
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
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (35)

Context-Inappropriate Capability

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

The fh function selects a historical command and executes it via eval, which can run arbitrary previously entered shell code with full user privileges. History often contains destructive, context-dependent, or attacker-influenced commands, so turning search into immediate execution materially increases the risk of unintended command execution.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

The guide presents eval-based execution of selected history commands without warning that the chosen entry may be harmful, stale, or unexpectedly expanded by the shell. This creates an unnecessary command-injection and accidental-execution risk because the selected text is treated as executable code rather than inert data.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The fkill workflow gives interactive destructive process-control capability and defaults to SIGKILL (-9), which abruptly terminates processes without cleanup. Embedding this in a fuzzy-finder guide normalizes a high-impact administrative action that can kill critical services or data-processing jobs with minimal friction.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The Docker workflows provide container administration actions including interactive exec into containers and stopping services, which exceeds the core purpose of a fuzzy finder and can directly affect running workloads. In an AI-assisted context, these shortcuts reduce friction for impactful actions on production-like environments without requiring explicit safety checks.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

High
Category
Not specified by scanner
Confidence
96% confidence
Finding

The embedded script includes file deletion capability, which is destructive and unrelated to the core fuzzy-selection behavior being documented. While it includes a confirmation choice, it still presents deletion as a ready-to-copy pattern that an AI might deploy or adapt inappropriately.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The skill advertises process management with fuzzy search and termination but provides no warning that terminating processes is destructive and may kill important services or lose unsaved work. In a tool-guiding AI workflow, presenting process killing as a normal capability without caution increases the chance of harmful operator mistakes.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The skill states that the AI will automatically install fzf and configure shell integrations, including generating reusable functions in shell configuration files, but it does not warn that these actions modify the user's system and persistent shell startup files. In an agent context, silent package installation or shell rc/profile edits can create unintended persistence, break user environments, or normalize unsafe consent-less system modification.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The instructions include downloading a binary and moving it into /usr/local/bin with sudo, which is a system-wide privileged change, but the guide does not clearly warn about the scope or trust implications. In an AI skill context, this is risky because it combines network retrieval with privileged installation into a shared executable path, potentially affecting all users and future command resolution.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The guide explicitly tells an AI to modify persistent shell startup files like ~/.bashrc and ~/.zshrc, which changes future user sessions, but it does not clearly warn that this is a lasting configuration change. In an agent-execution context, automatic edits to login scripts are security-sensitive because they can silently alter command behavior, persist mistakes, and create a foothold for further unintended execution whenever a shell starts.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The guide includes kill -9 $(ps aux | fzf | awk '{print $2}'), which enables force-terminating an arbitrarily selected process without validation, confirmation, or safer escalation. In a quickstart document for a fuzzy finder, this normalizes destructive system actions and increases the chance of accidental service disruption or data loss if a user or downstream agent executes it blindly.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

Demonstrating kill -9 without any warning is dangerous because SIGKILL bypasses graceful shutdown, preventing cleanup and increasing the risk of corruption or lost work. Presented in a quickstart, it may be copied verbatim by users or agents, making accidental termination of important processes more likely.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The guide explicitly instructs an AI to automatically modify a user's shell configuration, creating persistence and changing future command behavior without an explicit consent step. Although framed as convenience documentation, this crosses from fuzzy-finder usage into system modification and can silently introduce risky functions into the user's environment.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

Telling an AI to automatically add shell functions to startup files encourages persistent system changes without notifying or confirming with the user. Persistence in shell config is especially risky because it affects all future sessions and can silently alter command behavior long after the initial interaction.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

Defaulting to kill -9 teaches the most forceful termination method without explaining that it bypasses graceful shutdown and may corrupt state or lose data. The absence of warnings or a safer default makes accidental service interruption more likely.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The Git examples go beyond search and perform state-changing operations such as checkout, add, stash pop/drop, and branch creation. These are legitimate developer tasks, but in an AI-consumable skill they can lead to silent repository state changes, data loss, or confusion if executed automatically or without warnings.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The fco example reverts modified files using git checkout without clearly stating that uncommitted local changes will be discarded. In an AI-driven or copy-paste workflow, this omission can lead users to unintentionally destroy work they expected only to inspect or select.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The stash workflow performs drop and pop, both of which alter stash state and working tree contents, but the guide does not clearly warn about those side effects. Users may interpret the interface as read-only browsing and accidentally remove recovery points or reapply changes into an unsafe context.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The container stop example allows immediate interruption of selected services without warning that workloads, sessions, or dependent systems may be affected. In a fuzzy multi-select UI, a small selection error can cause outsized operational disruption.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The title and all troubleshooting instructions are written entirely in Chinese, with no indication that other languages are supported or that Chinese is a documented requirement. Under the policy, language constraints should be optional or clearly justified, and this file provides neither.

Content

No source excerpt is available for this finding.

Sudo/Root Execution

Medium
Category
Privilege Escalation
Confidence
70% confidence
Finding

Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.

Content

Scanner excerpt · guides/01-installation.md (reported line 88)May include surrounding context.

md
brew install fzf

# Ubuntu/Debian
sudo apt install fzf

# Arch Linux
sudo pacman -S fzf

Sudo/Root Execution

Medium
Category
Privilege Escalation
Confidence
70% confidence
Finding

Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.

Content

Scanner excerpt · guides/01-installation.md (reported line 107)May include surrounding context.

md
brew install fzf

# Ubuntu/Debian
sudo apt install fzf

# Arch Linux
sudo pacman -S fzf

Sudo/Root Execution

Medium
Category
Privilege Escalation
Confidence
70% confidence
Finding

Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.

Content

Scanner excerpt · guides/01-installation.md (reported line 110)May include surrounding context.

md
brew install fzf

# Ubuntu/Debian
sudo apt install fzf

# Arch Linux
sudo pacman -S fzf

Sudo/Root Execution

Medium
Category
Privilege Escalation
Confidence
70% confidence
Finding

Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.

Content

Scanner excerpt · guides/01-installation.md (reported line 113)May include surrounding context.

md
brew install fzf

# Ubuntu/Debian
sudo apt install fzf

# Arch Linux
sudo pacman -S fzf

Sudo/Root Execution

Medium
Category
Privilege Escalation
Confidence
70% confidence
Finding

Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.

Content

Scanner excerpt · guides/01-installation.md (reported line 116)May include surrounding context.

md
brew install fzf

# Ubuntu/Debian
sudo apt install fzf

# Arch Linux
sudo pacman -S fzf

Sudo/Root Execution

Medium
Category
Privilege Escalation
Confidence
70% confidence
Finding

Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.

Content

Scanner excerpt · guides/01-installation.md (reported line 232)May include surrounding context.

md
brew install fzf

# Ubuntu/Debian
sudo apt install fzf

# Arch Linux
sudo pacman -S fzf