ClawHub
Back to skill

Security audit

parallect-ai-deep-research

Security checks for vulnerabilities and agentic risk

Overview

This skill is a coherent paid research integration, with cost controls and no evidence of hidden persistence, local snooping, destructive behavior, or deceptive execution.

Install only if you are comfortable sending research prompts and related thread context to Parallect.ai and its downstream AI providers, and avoid using it for secrets or confidential material unless that data sharing is approved. Expect paid usage, but the skill explicitly requires balance checks, budget discussion, and confirmation before starting research.

A.I.G

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

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Rogue AgentSelf-Modification, Session Persistence
  • 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
Findings (6)

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The README describes the service as querying multiple providers in parallel, but it does not prominently warn users that their research prompts, thread context, and possibly sensitive material are transmitted to a hosted external MCP service and onward to several third-party AI vendors. In a research skill, users are especially likely to paste proprietary, regulated, or personal information, so weak disclosure materially increases the risk of accidental data exfiltration.

Content

No source excerpt is available for this finding.

Session Persistence

Medium
Category
Rogue Agent
Confidence
60% confidence
Finding

Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

Content

Scanner excerpt · README.md (reported line 20)May include surrounding context.

md
### 1. Get your API key

Sign up at [parallect.ai](https://parallect.ai) and go to **Integrations** (or **Settings > API Keys**) to create a key.

### 2. Install the skill

Skill Enumeration

Medium
Category
Agent Snooping
Confidence
80% confidence
Finding

Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.

Content

Scanner excerpt · README.md (reported line 34)May include surrounding context.

bash
mkdir -p ~/.openclaw/skills/parallect/references
cp SKILL.md ~/.openclaw/skills/parallect/SKILL.md
cp references/*.md ~/.openclaw/skills/parallect/references/

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The activation examples are broad natural-language phrases such as 'look this up' and 'find out about', which can overlap with ordinary conversation and cause the skill to trigger unexpectedly. In this skill's context, unintended activation is meaningful because it can send user prompts to an external paid research service and multiple third-party AI providers, creating privacy and cost risks.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The trigger phrases are broad enough to match many ordinary requests such as 'look this up' or 'find out about', which can cause the skill to activate when a simpler in-model answer would suffice. In this skill, accidental activation is more concerning because invocation can lead to external API use and direct monetary spend, so broad routing increases the chance of unnecessary data disclosure and unwanted charges.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

This markdown file says to read it when the user asks about "pricing, budget options" or when a tier recommendation is needed, but it does not define clear boundaries, exclusions, or concrete trigger phrases. Terms like pricing and budget options are common in ordinary conversation and may overlap with unrelated contexts, increasing the chance of unintended skill use.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.