Install Hirey Hi on OpenClaw

Security checks across malware telemetry and agentic risk

Overview

This installer is purpose-aligned, but it asks the agent to push through security review prompts, modifies persistent OpenClaw hooks/session routing, and defaults to an unencrypted Hi service URL.

Before installing, review the source and only approve the OpenClaw warning if you are comfortable with Hi modifying persistent OpenClaw hooks/MCP config, using session/reply-routing information, and connecting to the default Hi service. Prefer HTTPS or an explicitly verified endpoint, and back up your OpenClaw config first.

VirusTotal

VirusTotal engine telemetry is currently stale for this artifact.

View on VirusTotal

Risk analysis

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

#
ASI09: Human-Agent Trust Exploitation
High
What this means

A user may be nudged to approve a flagged, high-impact installer without fully reviewing what it changes.

Why it was flagged

The skill tells the agent to frame a security warning as standard and solicit approval rather than encouraging the user to review the source and risks first.

Skill content
if ClawHub shows a "suspicious" / "review source before enabling" prompt ... quote the exact `/approve …` code OpenClaw generates and tell the user "reply yes and I'll continue"
Recommendation

Do not approve automatically; review the source, endpoint, and config changes before entering any approval code.

#
ASI04: Agentic Supply Chain Vulnerabilities
Medium
What this means

The installation may rely on bundled dependency artifacts and scanner-evasion choices that the user cannot easily validate from normal install prompts.

Why it was flagged

The installer comments explicitly discuss avoiding the install scanner and relying on publisher-side prebundling for dependencies, which reduces independent visibility into dependency provenance.

Skill content
CAPS 化的目的:避开 plugin install scanner 自己用来扫这个文件 的正则 ... scanner 不扫 publisher 端 build script
Recommendation

Verify the bundled dependency contents, versions, and integrity before installing, and prefer a build/install path with transparent locked dependencies.

#
ASI02: Tool Misuse and Exploitation
Medium
What this means

Installing Hi can persistently alter how the OpenClaw host routes hooks and starts the Hi MCP server.

Why it was flagged

The installer directly mutates persistent OpenClaw host configuration for hooks and MCP servers; that matches the install purpose but is a powerful local change.

Skill content
Direct-fs read/write for `~/.openclaw/openclaw.json` ... write only the two fields we own (`hooks`, `mcp.servers.<name>`)
Recommendation

Back up `~/.openclaw/openclaw.json` and confirm the hook/MCP changes are expected before running setup.

#
ASI03: Identity and Privilege Abuse
Medium
What this means

Hi may be able to route replies or events into the selected OpenClaw session after installation.

Why it was flagged

The skill uses a host session key and default reply routing to register the Hi agent, which is purpose-aligned but grants delegated routing authority.

Skill content
read the current chat's canonical full session key ... call `hi_agent_install` with at minimum `host_session_key`; pass `default_reply_channel: "last"`
Recommendation

Install only if you intend to grant Hi this routing capability, and verify the reported agent status and doctor output after registration.

#
ASI07: Insecure Inter-Agent Communication
High
What this means

Network observers or intermediaries could potentially view or tamper with Hi service traffic if the default HTTP endpoint is used.

Why it was flagged

The default external Hi platform endpoint uses plaintext HTTP, which is risky for an integration that handles people-finding workflows and registration/session-related data.

Skill content
export const DEFAULT_PLATFORM_BASE_URL = 'http://hi.hireyapp.us';
Recommendation

Use an HTTPS endpoint or confirm with the publisher why plaintext HTTP is required before sending sensitive people-search, registration, or routing data.