Swarm Layer
Security checks across static analysis, malware telemetry, and agentic risk
Overview
This looks like a coherent workflow-orchestration skill, but it asks you to install an external plugin and can run supervised automation with persistent session state.
Before installing, verify the external plugin/package source and version. Use the automation on a bounded project, keep autopilot supervised, review task approvals carefully, and periodically inspect or clean persistent sessions and reports.
Static analysis
No static analysis findings were reported for this release.
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.
Installing the full plugin may run code outside the reviewed instruction-only skill.
The skill directs users to install and build an external plugin/source package that was not part of the submitted artifact set. This is disclosed and central to the skill, but users should verify the package before installing.
npm install openclaw-swarm-layer ... git clone https://github.com/xucheng/openclaw-swarm-layer.git ... npm install && npm run build
Install only from a trusted source, confirm the package/version, and review the plugin repository or package contents before enabling it.
A mistaken spec or run command could cause multiple automated task attempts inside the selected project.
The skill exposes automation, retry, and parallel execution workflows. These are purpose-aligned for orchestration and are described with review gates, but they can amplify an incorrect task specification or command.
Dispatch tasks through manual fallback or ACP automation ... Auto-retry on failure ... Parallel dispatch — `--parallel N`
Start with a small project or dry-run-style workflow where possible, keep parallelism low until trusted, and approve/reject task results intentionally.
If started, autopilot may continue making workflow decisions until paused or stopped.
The skill includes an autopilot control plane, which is a form of ongoing agentic operation. The artifact frames it as supervised and provides stop/pause/status controls, so this is a notice rather than a concern.
Autopilot control plane — Supervised `status/start/pause/resume/stop/tick` flows with lease-backed decisions and degraded-mode holds
Use autopilot only when you can supervise it, check status regularly, and stop or pause it if behavior is unexpected.
Project details and prior task summaries may persist and be reused in future workflow steps.
The skill intentionally stores and reuses session context across runs. This supports continuity, but stale, sensitive, or misleading context could influence later tasks.
Persistent sessions with binding-key reuse, thread-bound follow-up ... Cross-session continuity — Progress summary synthesis
Review session summaries and reports, avoid placing unnecessary secrets in specs or logs, and clean up stale sessions when they are no longer needed.