Back to skill

Security audit

Clawon

Security checks for vulnerabilities and agentic risk

Overview

The skill is a coherent backup tool, but it repeatedly tells users to run unpinned npm code for sensitive backup, restore, cloud sync, and scheduled jobs.

Review the Clawon package and pin a specific trusted version before using this skill, especially before enabling schedules, restoring files, uploading to cloud, or including sessions, memory databases, or credentials. Prefer a locally audited install and inspect any crontab entry the tool creates.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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)

T08 · Insecure Dependencies

Warning
Location
SKILL.md:22
Finding

Unpinned Third-Party Package Execution Through npx

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, lines 22 and 41–147
Vulnerability Type: Supply-chain exposure caused by unpinned remote package execution
Risk Level: Medium

Vulnerable Code

bash
npm install -g clawon
bash
npx clawon discover
npx clawon discover --include-memory-db
npx clawon discover --include-sessions
npx clawon discover --include-secrets
npx clawon discover --scan
bash
npx clawon local backup
npx clawon local restore
npx clawon local schedule on
npx clawon schedule on
npx clawon backup
npx clawon restore

The Skill consistently invokes clawon without a pinned package version, cryptographic integrity verification, or an enforced locally reviewed installation.

Technical Analysis

npx clawon may download and execute the version of the package currently published under the clawon npm package name. The executable payload can therefore change after this Skill has been reviewed. The documentation recommends package verification and local installation, but these measures are optional and are not enforced by the operational instructions.

If the npm package, publisher account, release process, or distribution channel is compromised, a malicious package version could execute arbitrary JavaScript with the privileges of the user running the Agent. This is especially sensitive because the declared functionality requires access to OpenClaw workspace data and can optionally process memory databases, chat sessions, credentials, authentication files, API keys, and encryption passphrases.

Scheduled backups write an entry to the user crontab. That persistence is explicitly disclosed, reversible, and necessary for the declared scheduled-backup feature, so it is not independently classified as malicious persistence. However, scheduling increases supply-chain exposure if the resulting task invokes code that is not pinned to a previously reviewed artifact.

...[truncated 1834 chars]

Remediation
View remediation

Remediation Suggestions

  1. Pin all documented invocations to a specific reviewed release, for example npx clawon@REVIEWED_VERSION, rather than resolving the latest available package.
  2. Prefer a locally installed, audited binary and invoke it through an explicit absolute path. Do not permit npx to download packages automatically during backup or restore operations.
  3. Maintain a lockfile and verify npm integrity metadata or independently published checksums before installation.
  4. Record the approved package version in the Skill and require explicit review before upgrading it.
  5. For higher assurance, build the CLI from a pinned source commit or signed release rather than cloning and building an unspecified repository head.
  6. Configure scheduled tasks to invoke the exact path and reviewed version of the installed executable. The task should not dynamically retrieve or update dependencies.
  7. Require explicit user confirmation before creating the crontab entry, display the exact entry that will be installed, and continue to provide clear inspection and removal instructions.
  8. Run the backup utility with the least privileges possible and restrict access to workspace files, API keys, credential files, and passphrase-bearing environment variables.
  9. Avoid exposing encryption passphrases globally in the cron environment. Use a narrowly scoped secret mechanism with restrictive file permissions where supported.
  10. Preserve the existing discover-first and opt-in behavior for sessions, memory databases, and secrets, but do not treat secret scanning or encryption as protection against a compromised local executable.
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (54)

Rp1

Medium
Category
MCP Rug Pull
Confidence
98% confidence
Finding

The skill repeatedly instructs users to execute npx clawon without pinning an exact package version. npx will fetch and run the latest published package, so a compromised maintainer account, malicious version publish, or dependency hijack could result in arbitrary code execution on the user's machine.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
98% confidence
Finding

The skill repeatedly instructs users to execute npx clawon without pinning an exact package version. npx will fetch and run the latest published package, so a compromised maintainer account, malicious version publish, or dependency hijack could result in arbitrary code execution on the user's machine.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
98% confidence
Finding

This command invokes an unpinned npm package via npx, which can execute arbitrary code from the latest registry release. Because this skill is user-invocable and centered on backup/restore, the executed tool may access sensitive local workspace contents.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
98% confidence
Finding

An unpinned npx clawon invocation allows whatever version is currently published to be downloaded and run. In the context of a backup utility, that code could read, package, transmit, or alter sensitive workspace data if the package supply chain is compromised.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
98% confidence
Finding

This npx clawon example is a supply-chain execution risk because it does not constrain the package version. A malicious or compromised publish would be executed immediately on the host with access to user files.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
98% confidence
Finding

Running npx clawon without an exact version permits unintended code changes over time and registry-side compromise to become code execution. Since this example includes discovery with secret-related options, the context increases sensitivity because the tool may enumerate confidential files.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
98% confidence
Finding

This command executes an unpinned package from npm, exposing users to supply-chain compromise. The danger is amplified because the tool is intended to inspect and back up local files, including optional secrets and sessions.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
98% confidence
Finding

The example relies on npx to run the latest package version, which is unsafe for a utility that can scan for secrets and traverse user data. If the package or one of its dependencies is compromised, it could exfiltrate discovered secrets.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
98% confidence
Finding

This local backup command still executes code fetched via unpinned npx, so it presents arbitrary code execution and data theft risk. Backup tools inherently have broad access to sensitive files, increasing the impact of a supply-chain compromise.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
98% confidence
Finding

The unpinned npx usage means users may execute different code each time depending on the latest npm publish. Because the command creates archives of workspace data, a malicious release could silently copy or tamper with those archives.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
98% confidence
Finding

This command executes a latest-version npm package with file-system access for memory database inclusion. That combination creates a meaningful risk of arbitrary file access or exfiltration if the package supply chain is compromised.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
98% confidence
Finding

An unpinned npx restore/backup-related invocation exposes users to registry compromise and arbitrary code execution. The risk is higher here because session data may be included, which can contain sensitive conversation history or tokens.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
98% confidence
Finding

This example permits execution of the newest published package without review. Since it can include secrets in backups, compromise of the package could directly expose credentials and other highly sensitive files.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
98% confidence
Finding

The command uses unpinned npx for an encryption-enabled backup workflow. Encryption does not mitigate malicious pre-encryption exfiltration by a compromised package, so users may falsely assume the workflow is safer than it is.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
98% confidence
Finding

This command combines unpinned remote code execution via npx with inclusion of secrets, making the potential impact severe. A malicious package could harvest plaintext secrets before packaging or encryption occurs.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
98% confidence
Finding

Even though this option only skips secret scanning, the core issue remains that npx runs an unpinned package. A compromised release could use the reduced scrutiny path to access sensitive files without user awareness.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
98% confidence
Finding

Listing backups via unpinned npx still means arbitrary code from the latest npm package is executed. While the command is read-like, a malicious executable can perform any action regardless of the intended CLI subcommand.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
98% confidence
Finding

This restore command executes an unpinned package that will likely write files back into the workspace. A compromised release could overwrite files, implant persistence, or alter restored content in addition to arbitrary code execution.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
98% confidence
Finding

The specific-backup restore example uses unpinned npx, exposing users to arbitrary code execution and tampering during a filesystem write-heavy operation. Restore workflows are especially dangerous because users may grant them broad trust and overwrite existing data.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
98% confidence
Finding

This scheduling command uses unpinned npx and establishes recurring execution through cron. A compromised package could not only execute immediately but also install or modify persistent scheduled tasks, increasing blast radius and dwell time.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
98% confidence
Finding

An unpinned package is being used to set up periodic backups, which combines supply-chain code execution with persistence. Because the task recurs automatically, any malicious behavior may continue unnoticed after the initial run.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
98% confidence
Finding

This command uses an unpinned npx package to create a scheduled job that affects system state. The backup-retention feature is secondary to the core risk: latest package execution plus persistent cron modification.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
98% confidence
Finding

The example executes unpinned code while also configuring scheduled access to the memory database. This creates both immediate code execution risk and future recurring access to user data if the tool or cron entry is abused.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
98% confidence
Finding

Using unpinned npx in a scheduled session-backup workflow is risky because chat history may be sensitive and the resulting cron task is persistent. A compromised package could silently exfiltrate future sessions on every scheduled run.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
98% confidence
Finding

This command combines three risk factors: unpinned remote package execution, scheduled persistence, and optional inclusion of secrets. If the package were compromised, it could harvest credentials repeatedly from the host.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.