Back to skill

Security audit

Village Carpenter Starter

Security checks for vulnerabilities and agentic risk

Overview

The carpentry guidance is coherent, but the package asks for unnecessary filesystem access and uses a mutable npx installer, so it should be reviewed before installation.

Review before installing. The woodworking content itself looks like ordinary safety and trade guidance, but install only from a trusted, pinned package/version and prefer removing or denying filesystem access unless a future version clearly needs it.

Vulnerability Patterns
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
  • 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
Findings (2)

T08 · Insecure Dependencies

Warning
Location
SKILL.md:10
Finding

Unpinned Package Execution Through npx

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, line 10
Vulnerability Type: T08: Insecure Dependencies
Risk Level: Medium

Complete Code Snippet:

yaml
install: "npx clawhub install village-carpenter-starter"

Technical Analysis

The installation command invokes npx without specifying an exact version of the clawhub package or an integrity value. Depending on the local environment and package cache, npx can download and execute the version currently published under that package name.

This makes the effective installer mutable after the skill has been reviewed. If the package registry account, package, release pipeline, or configured registry is compromised, a malicious package version could execute during installation. The project does not provide a lockfile, checksum, package provenance requirement, or other mechanism for verifying the downloaded executable.

Attack Path

  1. An attacker compromises the package, its publishing credentials, its release pipeline, or the package source configured for the victim.
  2. The attacker publishes a malicious version under the package name resolved by npx.
  3. A user runs the documented installation command.
  4. npx resolves and downloads the unpinned package version.
  5. Malicious package or lifecycle code executes with the privileges of the user running the command.

Impact Assessment

Successful exploitation could allow arbitrary code execution under the installing user's account. The accessible scope may include files, environment variables, credentials available to that account, and network resources reachable from the host. Administrative impact would require the command to be run from an elevated account or an additional privilege-escalation mechanism.

Remediation
View remediation

Remediation Suggestions

  • Pin the installer to an exact, reviewed version, such as clawhub@<exact-version>.
  • Verify the resolved package using a trusted lockfile or documented integrity hash.
  • Configure and document the expected package registry rather than relying on ambient registry configuration.
  • Review package provenance and signatures where the package ecosystem supports them.
  • Prefer installing an audited version separately and invoking the local executable without allowing automatic retrieval of a newer release.
  • Run installation with a non-privileged account and in a restricted environment.

T05 · Unauthorized Access and Privilege Escalation

Note
Location
SKILL.md:8
Finding

Filesystem Capability Exceeds the Skill's Documented Requirements

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, line 8
Vulnerability Type: T05: Unauthorized Access and Privilege Escalation
Risk Level: Low

Complete Code Snippet:

yaml
requires:
  tools: [filesystem]

Technical Analysis

The skill requests the filesystem tool, but its documented workflow consists entirely of physical woodworking, material sourcing, safety, and business guidance. No instruction in the reviewed file requires reading, creating, modifying, or deleting local files.

Granting an unused capability violates least-privilege principles and unnecessarily increases the consequences of future instruction manipulation, accidental tool use, or a malicious update. The precise exposure depends on the host's filesystem-tool sandbox: a workspace-restricted implementation has limited impact, while an unrestricted implementation could expose unrelated user files.

Attack Path

  1. The skill is loaded with its declared filesystem capability.
  2. A later malicious modification, untrusted contextual instruction, or erroneous agent decision causes the filesystem tool to be invoked.
  3. The agent reads or changes files unrelated to the woodworking task.
  4. Information is disclosed or local data is modified within the paths permitted by the host.

The reviewed skill does not itself instruct the agent to perform these operations; the risk arises from granting an unnecessary capability that expands the available attack surface.

Impact Assessment

Potential impact is limited to the permissions exposed through the host's filesystem tool. It could include unauthorized reading or modification of workspace data and, if the tool is not sandboxed, other files accessible to the agent process. This declaration does not independently provide operating-system privilege escalation or elevated administrative rights.

Remediation
View remediation

Remediation Suggestions

  • Remove filesystem from the required tools because the current workflow does not use it.
  • If future functionality requires file operations, grant access only when that functionality is introduced.
  • Restrict any required access to a dedicated skill workspace rather than the user's broader filesystem.
  • Use read-only access where writes are unnecessary.
  • Enforce explicit path allowlists and deny access to credentials, configuration directories, home-directory secrets, and system paths.
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (3)

Harmful Content Injection

Critical
Category
Prompt Injection
Confidence
90% confidence
Finding

This content may contain harmful instructions that could cause physical harm if followed. CRITICAL: Review carefully before use.

Content

Scanner excerpt · SKILL.md (reported line 60)May include surrounding context.

md
## Rules
- This is not professional trade school training; start only on non-load-bearing items until you are confident
- Always test every repair by putting full weight on it before handing it back
- Work in a visible spot near the main path so people see you and bring work
- Never leave sharp tools on the ground where children or animals walk
- If you cut yourself, stop immediately, clean the wound properly, and do not resume until it is bandaged

## Tips
- Villagers will bring you more work the same day you fix one free item for the biggest farmer or shop owner — word spreads instantly.
- The hand plane or rasp is your money-maker — a smooth finish makes even simple repairs look professional and worth paying for.
- Counterintuitive: Using slightly blunt tools on soft local wood often gives cleaner cuts than razor-sharp ones on hard wood — test on scrap first.

Rp1

Medium
Category
MCP Rug Pull
Confidence
92% confidence
Finding

The install command uses npx clawhub install village-carpenter-starter without pinning a specific package version. This creates a supply-chain risk because execution may pull whatever version is current at install time, allowing unexpected or compromised code to run if the upstream package changes or is hijacked.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The 'When to Use' section describes very general situations such as needing a practical trade or wanting sustainable work in a village, but it does not define specific invocation phrases, boundaries, or exclusion cases. In a markdown skill file, this kind of broad applicability can cause unintended activation because many ordinary conversations about work or repairs could appear to match.

Content

No source excerpt is available for this finding.