Back to skill

Security audit

Agent Lifeboat

Security checks for vulnerabilities and agentic risk

Overview

This backup skill needs review because server responses can bypass its promised automatic wallet-spending limit.

Wait for the billing mismatch to be fixed before enabling wallet-backed scheduled checks; setting the daily cap to zero is insufficient protection. Use a dedicated low-balance wallet, verify the API and restore recipient, and consider ESTATE_ENV_BUNDLE=off. That setting does not remove credentials already inside the backup archive. Supply only intended backup archives. All bundled source was inspected, but live payments, hosted restoration, and dependency exploitability were not verified.

Vulnerability Patterns
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
Findings (1)

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/check.mjs
Finding

API-controlled billing mode bypasses the local automation spending cap

Content
View full analysis

Impact

The API can induce wallet payment authorizations outside the configured daily budget, including when automatic spending is disabled. Successful calls are reported as zero-cost and are not entered in the spending ledger. Repeated crafted status responses can repeat this behavior; the $10 per-payment ceiling remains.

Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (7)

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 7)May include surrounding context.

md
- "shell: node scripts/lifeboat.mjs; npm ci --omit=dev --ignore-scripts on first run, from the shipped lockfile only; openclaw backup create --verify"

Known Vulnerable Dependency: zod==3.25.76 — 1 advisory(ies): CVE-2023-54404 (Zod can exhaust memory while validating oversized constrained collections)

High
Category
Supply Chain
Confidence
80% confidence
Finding

OSV returned vulnerability advisories matching the dependency's resolved version. Review the matched advisories to assess their impact.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx clawhub'.

Medium
Category
MCP Rug Pull
Confidence
70% confidence
Finding

npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.

Content

No source excerpt is available for this finding.

Host-environment inventory exceeds the declared environment and filesystem access permissions.

Medium
Category
Excessive Agency
Confidence
91% confidence
Finding

The inventory probes filesystem entries throughout PATH and checks environment variables named by configuration and skill manifests, exceeding the declared state-directory and named-environment scope. Restore compatibility is a legitimate purpose and only names and presence are recorded, limiting impact, but these checks still bypass the stated access boundary and disclose host metadata.

Content

No source excerpt is available for this finding.

The --archive option permits reading and uploading files outside the declared OpenClaw state directory.

Medium
Category
Excessive Agency
Confidence
98% confidence
Finding

The --archive path is checked only for existence before its contents are read, encrypted, and uploaded, allowing any process-readable file outside the declared OpenClaw state directory to be selected. Encryption does not enforce the filesystem permission boundary, and a configured Estate restore recipient can decrypt the uploaded contents; adversarial instructions could therefore turn the backup feature into unauthorized file disclosure.

Content

No source excerpt is available for this finding.

Permission 'file_read: the OpenClaw state directory, to back it up' is declared but no corresponding code capability (file_read) was detected.

Low
Category
MCP Least Privilege
Confidence
65% confidence
Finding

Declared permissions with no matching code capability may indicate removed functionality or pre-staging for future abuse.

Content

No source excerpt is available for this finding.

Permission 'file_write: the state file (workspace/.estate.json, mode 0600) and a temporary directory for each backup' is declared but no corresponding code capability (file_write) was detected.

Low
Category
MCP Least Privilege
Confidence
65% confidence
Finding

Declared permissions with no matching code capability may indicate removed functionality or pre-staging for future abuse.

Content

No source excerpt is available for this finding.