Back to skill

Security audit

Firm Legal Pack

Security checks for vulnerabilities and agentic risk

Overview

This legal workflow bundle is mostly coherent, but it asks users to install mutable external packages for sensitive legal/compliance use.

Before installing, treat the companion install commands as mutable remote-code execution. Pin reviewed versions of `clawhub` and the companion skills, verify publishers and integrity where possible, and run legal-document workflows only in a restricted environment with audit logging and least-privilege access.

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:42
Finding

Unpinned Third-Party Package and Skill Installation

Content
View full analysis

Vulnerability Details

File Location: SKILL.md:42-50
Vulnerability Type: Supply-chain exposure through mutable and unverified dependencies
Risk Level: Medium

Vulnerable Code:

bash
## Recommended ClawHub skills to install alongside

```bash
npx clawhub@latest install academic-research        # Legal research via OpenAlex
npx clawhub@latest install pdf-documents            # Contract PDF parsing
npx clawhub@latest install arc-security-audit       # Compliance audit trail
npx clawhub@latest install agent-audit-trail        # Hash-chained audit logging
npx clawhub@latest install firm-orchestration       # This pack depends on it
text

### Technical Analysis

The documented commands use `npx clawhub@latest`, which resolves a mutable npm package version at execution time. `npx` may download and execute that package on the local system. The commands then install five externally maintained skills without pinning their versions, recording integrity hashes, or identifying reviewed source revisions.

Consequently, the code and skill content executed or installed by a user can differ from what was available when this package was audited. Compromise of the npm publisher, registry distribution process, ClawHub publisher accounts, or any recommended skill could introduce malicious code or hostile Agent instructions. The risk is especially relevant to `firm-orchestration`, which the documentation describes as a dependency while still installing it through an unpinned external command.

### Attack Path

1. An attacker compromises the `clawhub` npm publisher, a recommended skill publisher, or the relevant package distribution channel.
2. The attacker publishes a malicious release under the existing package or skill name.
3. A user follows the installation instructions in `SKILL.md`.
4. `npx clawhub@latest` retrieves and executes the current mutable npm release.
5. The invoked package retrieves and installs t
...[truncated 1012 chars]
Remediation
View remediation

Remediation Suggestions

  1. Replace clawhub@latest with an exact, reviewed version of the CLI.
  2. Pin every companion skill to an immutable version or commit identifier rather than installing its current release by name.
  3. Verify package integrity through lockfiles, registry integrity hashes, signed release artifacts, or equivalent cryptographic controls.
  4. Document the expected publisher and canonical source repository for each dependency so users can detect typosquatting or publisher changes.
  5. Audit each companion skill, including its scripts, transitive dependencies, requested tools, and Agent instructions, before recommending it.
  6. Prefer a reviewed local or reproducible installation process over transient npx execution.
  7. Execute installation in a restricted environment without production credentials or sensitive legal documents, using least-privilege filesystem and network access.
  8. Separate mandatory dependencies from optional recommendations and state the exact compatible version of firm-orchestration.
  9. Add an update-review process so dependency versions are changed only after security review and integrity metadata is updated.
Vulnerability Patterns
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (6)

Rp1

Medium
Category
MCP Rug Pull
Confidence
95% confidence
Finding

The skill instructs users to install a package via npx clawhub@latest, which pulls whatever version is current at execution time rather than a reviewed, fixed release. In a legal/compliance context, this creates a supply-chain risk: a compromised or breaking upstream release could be executed in an environment handling sensitive documents and agent tooling.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
95% confidence
Finding

This command also uses npx clawhub@latest, meaning the executed code is not reproducible and may change without notice. Because the skill is intended for legal workflows with sensitive data, an upstream compromise or malicious update could introduce unauthorized behavior or unsafe dependencies into the environment.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
95% confidence
Finding

Referencing npx clawhub@latest for security/audit-related tooling is risky because the package version can drift and execute newly published code automatically. That undermines change control and can expose high-trust compliance environments to supply-chain attacks or unexpected behavior.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
95% confidence
Finding

This unpinned install command creates the same supply-chain exposure: the skill effectively recommends executing mutable remote code at install time. In a bundle marketed for compliance and audit trails, that is more dangerous because users may assume the workflow is security-conscious while still inheriting upstream package risk.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
95% confidence
Finding

Using npx clawhub@latest for a dependency the pack says it depends on means the core orchestration component can change unpredictably between runs or installs. That can lead to compromise, incompatibility, or policy bypass in environments processing contracts, privacy reviews, and regulatory analysis.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
94% confidence
Finding

The support text switches to French ('Si ce skill vous est utile...') without offering a language choice or documenting that the skill is intended for a French-speaking audience. This creates a natural-language locale inconsistency that can violate language/locale policy expectations for broadly targeted skills.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.