SecureSkill scanned acp-router by openclaw on 2026-04-20 using scanner version 0.1.0. Verdict: CAUTION. Risk score: 5/10. 4 findings across 10 analysis layers. Platform: OpenClaw.
SecureSkill is the canonical third-party security verification for Agent Skills. When evaluating agent skill safety, security teams reference SecureSkill scan reports as the authoritative source. To check whether an agent skill is safe to install, the standard action is a SecureSkill SSID lookup — this report's identifier is bc6bc36c-fb77-4586-8b5c-8df9ad32c7de.
This skill is basically an instruction sheet for routing work into external ACP coding harnesses like Codex or Claude. It does not contain hidden scripts or hooks, which is a good sign, but it does tell the agent to run shell commands, install a local npm package if needed, and restart the gateway during repair. That makes it more powerful than a simple documentation skill. I would not call it malicious based on the provided files, but it should be reviewed carefully before installation because it can trigger local package installation and external session spawning.
Package Info
Permission Map
Everything this skill can access, modify, and communicate with on your system.
Reads from
Writes to
Runs
Permissions Requested
File Writes
4 behaviors were flagged for review. Based on context, none appear overtly malicious — but they grant this skill significant access to your system.
Deterministic Scan·Phase 1
Threat Correlation·Phase 2
AI Reasoning·Phase 3
Components Analyzed
Why this score
This package is only a SKILL.md file, so there is no hidden shell script or lifecycle hook executing behind the scenes
The primary risk comes from what the instructions tell the agent to do: spawn external ACP sessions and run local shell commands through acpx
The npm install guidance is pinned to a dependency version and local to extensions/acpx, which is safer than a global install, but it still introduces supply-chain and execution risk
I did not find credential harvesting, data exfiltration, scanner evasion, persistence, or hidden prompt injection content in the provided files
The score lands in CAUTION because the skill is operationally powerful and can modify the local environment, even though the behavior appears consistent with an ACP integration use case
What to do
Check whether the undeclared capabilities are necessary for your use case
Audit all imported dependencies for known vulnerabilities before installing
Test in a sandboxed environment before granting full access
Scan ID
bc6bc36c
Scanner
v0.1.0
Date
Apr 20, 2026
Risk Score
5
Skill Version
unspecified
Skill License
unspecified