Skill flagged — suspicious patterns detected

ClawHub Security flagged this skill as suspicious. Review the scan results before using.

Claw Time Machine

v1.0.4

Backup, restore, and migrate OpenClaw installations. Preserve workspace memories, credentials, custom skills, scheduled tasks, and core configuration. Use wh...

1· 72·0 current·0 all-time
byTacit Lab@tacitlab
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
VirusTotalVirusTotal
Pending
View report →
OpenClawOpenClaw
Suspicious
medium confidence
Purpose & Capability
Name and description match the script behavior: it collects and archives OpenClaw state (workspace, credentials, skills, etc.), lists backups, restores them, and can migrate via scp/ssh. However the registry metadata declares no required binaries or environment variables while the script clearly depends on standard CLI tools (tar, ssh, scp, cp, mktemp, du, stat, find, sort). That metadata omission is an inconsistency (likely sloppy packaging) but reduces transparency.
Instruction Scope
SKILL.md instructs the agent to run the bundled script and documents backup/list/restore/migrate flows. The script's actions (copying specific OpenClaw state paths, creating manifests, creating a safety backup, using scp/ssh to run a remote restore script) are within the stated scope. The instructions do not ask the agent to read unrelated system areas, but they do operate on sensitive OpenClaw files (credentials, telegram, identity) which is expected for this tool.
Install Mechanism
No install spec is provided and the skill is delivered as an instruction plus a script file. That is the lowest install-risk category. Nothing in the package pulls arbitrary remote binaries or runs external installers.
Credentials
The skill requests no environment variables or explicit credentials in metadata, yet it archives and restores sensitive directories (credentials/, telegram/, identity/) and performs remote copy/restore over SSH. These actions legitimately require access to sensitive files and an SSH connection to the target host, but the metadata does not call this out. Users should recognize that backups will include credentials and that migration pushes those credentials to the remote host.
Persistence & Privilege
The skill is not marked always:true and does not request elevated platform privileges. Autonomous invocation (disable-model-invocation=false) is default; there is no other unusual persistence or cross-skill configuration modification.
What to consider before installing
This skill appears to implement the advertised backup/restore/migrate functionality, but take these precautions before installing or using it: - Review the full scripts yourself. The package includes scripts/ctm.sh — inspect it end-to-end to confirm there is no hidden network exfiltration or unexpected commands. The manifest you were shown is truncated in the copy you received; request the complete file if unsure. - Understand what will be backed up: credentials/, telegram/, identity/ contain sensitive secrets. Backups include those by design; do not migrate them to untrusted hosts. - Migration uses scp and runs a restore script over ssh on the target host. Only migrate to machines you control or fully trust and ensure your SSH keys are protected. The skill does not perform remote authentication hardening. - The package metadata omits required binaries (tar, ssh, scp, cp, mktemp, du, stat, find, sort). Expect the script to fail if those tools are missing; this metadata mismatch is sloppy and reduces transparency. - Test in a disposable environment first: run backup and restore locally with --dry-run or on a throwaway VM to ensure behavior matches expectations and safety backups are created. - If you need higher assurance, ask the publisher for: (1) a full, untruncated copy of scripts/ctm.sh; (2) an explicit statement about what files inside credentials/ and identity/ are copied; (3) a signed release or source repo so you can audit history. Given the metadata inconsistencies and the sensitive nature of data being handled, proceed only after you review the script and confirm you trust the target host(s).

Like a lobster shell, security has layers — review code before you run it.

latestvk975g66bcvw4xkq7xr2ne9zt01843040

License

MIT-0
Free to use, modify, and redistribute. No attribution required.

Comments