Back to skill

Security audit

Collabs & Cross-Promotion

Security checks for vulnerabilities and agentic risk

Overview

This is a text-only planning skill for creator collaborations, with some confusing giveaway and native-collab examples that users should review before relying on it.

Installers should understand this skill is advisory content, not executable automation. Review outputs carefully when a request involves giveaways, paid/free-product exchanges, native IG/TikTok/YouTube collaboration features, or public posting, because the artifact contains some mixed guidance about what WoopSocial can plan versus what must remain native or be handled by another workflow.

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
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (6)

Intent-Code Divergence

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

The worked example gives a concrete joint giveaway workflow despite the manifest stating the skill does not run giveaways or manage the agreement. Because examples are strong behavioral priors for agents, this can directly induce the model to facilitate a prohibited promotion format and provide incorrect disclosure guidance, creating legal/compliance exposure and user harm.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The reference file recommends unsupported formats such as joint giveaways and native collaboration-style posting even though the skill metadata explicitly says WoopSocial does not support giveaways, duet/stitch, or native IG Collab co-authored posts. This mismatch can cause the agent to generate unsafe or impossible instructions, leading users into policy, compliance, or platform-operation errors because the reference content overrides product boundaries.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The file includes joint giveaways as a recommended collaboration format even though the manifest explicitly says the skill does not run giveaways. An agent using this reference may incorrectly propose or structure giveaway campaigns, pushing the user into unsupported legal, disclosure, and platform-rule-sensitive activity.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The reference text instructs use of native platform collaboration features and formats such as IG Collab posts, TikTok duet/stitch, and YouTube's native Collaboration feature, while the skill metadata explicitly states this skill does not create native IG Collab posts, duet/stitch, or manage those platform-native workflows. This mismatch can cause the agent to recommend or attempt unsupported actions, producing unsafe task drift, user confusion, or policy-noncompliant outputs outside the skill's declared scope.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
87% confidence
Finding

The file says that if money or free product changes hands, the matter should be handled via ugc-and-influencer or creator-monetization. The manifest states this skill is distinct from creator-monetization and that paid/free-product arrangements make it a sponsorship routed to ugc-and-influencer; mentioning creator-monetization here blurs that boundary and contradicts the stated intent separation.

Content

No source excerpt is available for this finding.

Vague Triggers

Low
Category
Not specified by scanner
Confidence
91% confidence
Finding

The eval inputs use very broad, natural-language triggers like 'Help me set up a collab' and 'Plan a cross-promo partnership for us,' which can overlap with ordinary user requests outside a tightly scoped collab workflow. This can cause the skill to activate too readily or be selected over a more appropriate sibling skill, increasing the chance of incorrect routing or unintended handling of adjacent tasks such as paid influencer deals or DM outreach.

Content

No source excerpt is available for this finding.