Trusted — Risk Score 0/100
Last scan:1 day ago Rescan
0 /100
openclaw-hi-install
Installs Hi into a local OpenClaw host through the official ClawHub path, then completes register, activate, receiver setup, and health checks through Hi's high-level install control tools.
This skill is purely documentation for installing the Hi agent platform on OpenClaw hosts. No code or scripts are present—only a comprehensive SKILL.md describing an installation workflow with appropriate security practices.
Skill Nameopenclaw-hi-install
Duration27.5s
Enginepi
Safe to install
This skill is safe to use. However, the actual security depends on the remote npm packages (`@hirey/[email protected]`, `@hirey/[email protected]`) and the `hi_agent_install`/`hi_agent_doctor` commands it references. Ensure these dependencies are from a trusted source before execution.
ResourceDeclaredInferredStatusEvidence
Filesystem WRITE WRITE ✓ Aligned SKILL.md: Documents writes to ~/.openclaw/vendor/hi for user-local npm installat…
Network READ READ ✓ Aligned SKILL.md: Documents configuring OpenClaw hooks at http://127.0.0.1:18789/hooks/a…
Shell WRITE WRITE ✓ Aligned SKILL.md: Documents npm install commands and hi_agent_* tool invocations
Environment WRITE WRITE ✓ Aligned SKILL.md: Documents setting HI_PLATFORM_BASE_URL, HI_MCP_TRANSPORT, HI_MCP_PROFI…
Skill Invoke READ READ ✓ Aligned SKILL.md: Invokes hi_agent_install, hi_agent_doctor, hi_agent_reset tools
1 findings
🔗
Medium External URL 外部 URL
http://127.0.0.1:18789/hooks/agent
SKILL.md:35

File Tree

1 files · 7.2 KB · 87 lines
Markdown 1f · 87L
└─ 📝 SKILL.md Markdown 87L · 7.2 KB

Dependencies 2 items

PackageVersionSourceKnown VulnsNotes
@hirey/hi-mcp-server 0.1.6 npm No Version pinned; trust depends on @hirey namespace
@hirey/hi-agent-receiver 0.1.7 npm No Version pinned; trust depends on @hirey namespace

Security Positives

✓ No executable code present—pure documentation
✓ Uses version-pinned npm packages (0.1.6, 0.1.7) to prevent supply chain attacks
✓ Installs to user-local directory (~/.openclaw/vendor/hi) instead of requiring elevated privileges
✓ Generates fresh random tokens for OpenClaw hooks (does not reuse gateway auth token)
✓ Has explicit security boundaries: 'do not ask for AWS credentials'
✓ Requires user confirmation before permission/auth prompts or destructive operations
✓ Documents proper validation steps with hi_agent_doctor
✓ Explicitly warns against using raw-skill or ad-hoc install paths