Destructive delete command
Warn
- Finding
- Documentation contains a destructive delete command without an explicit confirmation gate.
AgentWallex Payment
Security checks across static analysis, malware telemetry, and agentic risk
Overview
The skill is clearly about AgentWallex payments, but it gives the agent high-impact payment and account-management power and only clearly gates pay/transfer actions, not all destructive account changes.
Install only if you are comfortable letting your agent access AgentWallex. Start in sandbox, use the least-privileged key available, verify any production payment details carefully, and require confirmation before any create/update/suspend/delete action as well as payments.
Static analysis
VirusTotal
VirusTotal findings are pending for this skill version.
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.
What this means
An agent could potentially update, suspend, or delete AgentWallex agents without an explicit confirmation workflow, which may disrupt wallets or payment operations.
Why it was flagged
The skill documents mutating AgentWallex account actions, but only transfer/pay are explicitly named for human confirmation while autonomous model invocation is enabled.
Skill content
"disableModelInvocation": false, "humanConfirmation": {"required": true, "actions": ["transfer", "pay"]} ... `agentwallex_delete_agent` | Yes | Delete an agent | ... `agentwallex_agent_status` | Yes | Update agent status (active / suspended)Recommendation
Require explicit user confirmation for every mutating action, including create, update, suspend, and delete; document whether those actions are reversible and what funds or wallets they affect.
What this means
If the local config file or pasted API key is exposed, someone may be able to use the associated AgentWallex account permissions.
Why it was flagged
The skill handles and persists an AgentWallex API key; this is expected for the integration, but it is a high-value credential for payment/account operations.
Skill content
No API key or environment variable is needed upfront — credentials are configured through conversation and stored locally. ... Credentials are stored locally at `~/.openclaw/agentwallex/config.json` with owner-only permissions (0600).
Recommendation
Use sandbox first, prefer least-privilege or limited production keys if available, avoid pasting keys into untrusted sessions, and rotate the key if the config file is exposed.
What this means
Running the command will delete saved AgentWallex credentials and local configuration for this skill.
Why it was flagged
The shell delete command is in user-directed uninstall documentation and is scoped to the skill's own local configuration directory.
Skill content
To also remove locally stored credentials: ```bash rm -rf ~/.openclaw/agentwallex ```
Recommendation
Run this command only when you intentionally want to remove the skill's stored credentials; consider backing up needed configuration first.