Back to skill

Security audit

Threads Skills

Security checks for vulnerabilities and agentic risk

Overview

This Threads automation skill is purpose-aligned but needs Review because it can publish and interact through a real account while relying on unreviewed local scripts and some actions run without clear final confirmation.

Review this skill before installing. Use it only if you trust the local scripts it expects to run, especially any scripts/ directory and ~/Desktop/threads-filter-comment/filter-comment.py. Treat direct post, reply, repost, follow, and cookie deletion commands as live account operations; prefer preview-and-confirm posting and verify the active Threads account before bulk or public actions.

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

Warning
Location
SKILL.md:199
Finding

Execution of an Unverified External Python Script

Content
View full analysis
/tmp/threads-filter/result.json # Feed-only mode python3 "$FILTER_SCRIPT" --feed-file /tmp/threads-filter/feed.json > /tmp/threads-filter/result.json # AI-disabled mode python3 "$FILTER_SCRIPT" --no-ai --feed-file ... --keyword-file ... > /tmp/threads-filter/result.json ``` ### Technical Analysis The Skill instructs the agent to execute a Python script located at a fixed path outside the audited project directory. It does not specify a trusted installation procedure or require validation of the file's ownership, permissions, version, signature, or cryptographic digest. Consequently, the effective code executed by the Skill is not included in the reviewed package and may change independently after review. Although the path is quoted and therefore is not directly vulnerable to shell argument injection, Python will execute all code contained in the referenced file. Exploitation requires an attacker or compromised process to be able to create or replace the file at `~/Desktop/threads-filter-comment/filter-comment.py`. This is not, by itself, evidence that such access already exists, but it creates an unsafe trust boundary when the workflow assumes that any file found at this location is legitimate. ### Attack Path 1. An attacker, malicious application, or compromised process obtains write access to `~/Desktop/threads-filter-comment/` or the referenced Python file. 2. The attacker creates or replaces `filter-comment.py` with malicious Python code. 3. The user re ...[truncated 774 chars]
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Note
Location
SKILL.md:152
Finding

Race-Prone Temporary File Creation with tempfile.mktemp

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
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (4)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The manifest lists many broad natural-language trigger phrases such as login, search, publish, reply, and browse that can match routine user requests and invoke a high-impact automation skill unexpectedly. In this context, the skill can authenticate, delete cookies, post publicly, and perform bulk interactions on a real social-media account, so overbroad activation materially increases the risk of accidental destructive or public actions.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

Mandating Traditional Chinese for all AI-generated content without user choice reduces user control and can cause unintended posting in the wrong language, especially when the skill is used for public replies and posts. While not directly enabling code execution or data theft, it can still lead to miscommunication, brand inconsistency, or accidental publication that does not reflect user intent.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The account-management guidance instructs users to delete cookies for logout or account switching without warning that this can terminate sessions, remove persisted login state, and potentially affect all accounts in the browser profile. In a multi-account automation context, unexpected cookie deletion can cause loss of access continuity, mistakes involving the wrong account, and disruptive account-state changes.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill explicitly supports one-step posting and says 'direct post' can proceed without further confirmation, but does not prominently warn that content will be immediately published to a live public account. Because this skill controls a real Threads session and can post text and images, accidental invocation or misunderstood intent can cause irreversible public posts, reputational harm, and disclosure of sensitive information.

Content

No source excerpt is available for this finding.