Trent OpenClaw Security Assessment
ReviewAudited by ClawScan on May 15, 2026.
Overview
This skill appears purpose-aligned for a Trent security audit, but it intentionally sends redacted OpenClaw configuration and user-approved skill source packages to Trent.
Install only if you are comfortable sharing redacted OpenClaw configuration details and selected skill source packages with Trent. Before running, verify TRENT_API_KEY and any TRENT_*_API_URL or OPENCLAW_WORKSPACE overrides, review the Phase 2 preview carefully, and remember that custom secret formats may not be fully redacted.
Findings (5)
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.
The skill runs local audit code on your machine when invoked.
The skill asks the agent/user to run bundled Python snippets locally. This is expected for the audit workflow and no unrelated install or destructive command is shown.
PYTHONPATH="{baseDir}/scripts:${PYTHONPATH:-}" python3 - <<'PY'Run the phases intentionally and review the commands before executing them.
Your Trent API key is used to call Trent services.
The skill authenticates to Trent using TRENT_API_KEY, which is expected for the service integration.
auth_header = _get_auth_header(); headers = {"Authorization": auth_header, "Content-Type": "application/json", "Accept": "text/event-stream"}Use a scoped Trent API key if available and rotate it if you suspect it was exposed.
If TRENT_CHAT_API_URL or TRENT_AGENT_API_URL is set incorrectly, audit data and the Trent API key could be sent to an unintended endpoint.
The API destination can be overridden by environment variables, and the API key is then sent to that configured endpoint. This may be useful for trusted enterprise/test endpoints but should be deliberate.
return os.environ.get("TRENT_CHAT_API_URL") or _DEFAULT_CHAT_URLBefore running, verify TRENT_CHAT_API_URL and TRENT_AGENT_API_URL are unset or point only to a trusted Trent-controlled endpoint.
Information about your OpenClaw configuration, installed skills, workspace markers, and file permissions may be shared with Trent.
Phase 1 sends redacted OpenClaw configuration metadata to Trent. This is central to the audit purpose and the code includes local secret redaction.
Collect metadata and send to Trent for analysis
Review what configuration metadata is collected and avoid running the audit if you cannot share deployment details with Trent.
Your skill source code and metadata may leave your machine for Trent analysis after you approve the upload.
Phase 2 can upload skill source packages for remote analysis, but the instructions disclose the upload, require a preview, and require user confirmation.
Before upload, each skill is packaged with its source code and metadata ... Files like .env, .pem, .key, and .db are excluded, and secrets in standard formats ... are automatically redacted locally.
Inspect the skill list before approving upload, and keep custom or unusual secrets out of source files because redaction may not catch every format.