ClawHub
Back to skill

Security audit

HCTI Open Graph Images

Security checks for vulnerabilities and agentic risk

Overview

This skill gives clear instructions for setting up HCTI Open Graph image rendering and does not include hidden code, automatic credential access, or persistence beyond the expected user-authorized MCP connection and website configuration.

Before installing, confirm you want to connect your OpenClaw client to HCTI and that you are comfortable sending the relevant page HTML/CSS, URLs, template values, and render options to HCTI. Review account costs and image-credit usage, and only allow proxy, storage-destination, update, or delete actions when you explicitly intend those account changes.

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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (1)

Credential Access

High
Category
Privilege Escalation
Content
Have the user complete browser authorization and choose their HCTI organization. Do not overwrite an existing `hcti` entry pointing elsewhere. For an existing correct connection, run `openclaw mcp login hcti` to authorize it. Check connectivity with `openclaw mcp doctor hcti --probe`, then ensure the current agent session exposes the HCTI tools before rendering. If it does not, reload the MCP connection or start a new session as supported by the client.

Installing this skill supplies instructions; it does not automatically register or authorize the MCP server. Use the client's OAuth credential storage, and never copy access tokens into skill files, shell commands, or chat. If the installed OpenClaw version lacks these MCP commands, consult its supported connection setup or update it before proceeding.

HTML, CSS, target URLs, template values, and rendering options are processed by the hosted HCTI service. See [HCTI MCP documentation](https://docs.htmlcsstoimage.com/integrations/mcp/) and [OpenClaw MCP setup](https://docs.openclaw.ai/cli/mcp/registry).
Confidence
70% confidence
Finding
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Static analysis

No suspicious patterns detected.