Catchclaw Agentar

Security checks across static analysis, malware telemetry, and agentic risk

Overview

The skill appears to be a coherent CatchClaw package manager, but it can run a bundled Node CLI that downloads, installs, overwrites, exports, and rolls back OpenClaw agent workspaces.

This skill looks purpose-aligned, but treat it like a package manager for agents: install only trusted agentars, choose “new” rather than “overwrite” unless you truly want to replace the main workspace, be careful with API keys, and review ZIP exports before sharing them.

Static analysis

No static analysis findings were reported for this release.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

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

A mistaken overwrite could replace the user's active agent instructions, skills, and configuration, though the skill says it creates a backup.

Why it was flagged

The install command can overwrite the main agent workspace, which is high-impact. The artifact also includes an explicit user-selection requirement and backup behavior, making it disclosed and purpose-aligned rather than a concern.

Skill content
$CLI install <slug> --overwrite ... Overwrite the main agent (`~/.openclaw/workspace`). Existing workspace is backed up automatically. **Never use without the user's explicit selection.**
Recommendation

Prefer installing to a new named agent unless the user clearly wants to replace the main workspace, and confirm the exact slug and mode before running install.

What this means

Installing an untrusted agentar could change how an agent behaves in future sessions.

Why it was flagged

Installing marketplace archives is the core purpose, but those archives can introduce new agent instructions, skills, and configuration. The visible artifacts show ZIP validation and clear user-facing commands, but no signature or trust verification mechanism is described.

Skill content
An agentar is a distributable agent archive (ZIP) containing workspace files such as SOUL.md, skills, and other configuration. It can be installed as a new agent or used to overwrite an existing agent with a single command.
Recommendation

Install only agentars from sources you trust, prefer a new workspace first, and review the installed files before using them for sensitive work.

What this means

Running the skill executes local Node code and may invoke the local OpenClaw CLI as part of workspace management.

Why it was flagged

The bundled CLI can spawn the local OpenClaw binary. The code comments and shell:false setting indicate an attempt to bound this behavior, and it fits the purpose, but it is still local command execution.

Skill content
const _cp = builtinModules.find(m => m.length === 13 && m[5] === '_' && m.startsWith('c'));
const { spawnSync } = _require(`node:${_cp}`);
...
return spawnSync(openclawBin, args, { ...options, shell: false });
Recommendation

Use the skill only from a trusted installation path and run it under a normal user account, not with elevated privileges.

What this means

A provided API key may remain on disk and could be exposed if the workspace or credentials file is shared or backed up insecurely.

Why it was flagged

The skill can store a user-provided API key locally for installed agentars. This is disclosed and optional, but it is credential handling.

Skill content
`--api-key <key>` — (Optional) API key to save into `skills/.credentials` for agentars that require backend authentication.
Recommendation

Provide API keys only when needed, use least-privileged keys, and avoid sharing exported workspaces that may contain credential files.

What this means

If memory is included in an export, private notes or long-term agent context could be packaged into a ZIP that may later be shared.

Why it was flagged

The export function can package persistent memory if the user explicitly chooses --include-memory. The default excludes memory and filters common secret filenames, which reduces but does not eliminate sensitive-data risk.

Skill content
$CLI export [--agent <id>] [-o <path>] [--include-memory] ... MEMORY.md is excluded by default ... `--include-memory` — Include MEMORY.md in export
Recommendation

Leave memory excluded unless you intentionally want to package it, and inspect exports before distributing them.