Back to skill

Security audit

KingOfZhao Decision Framework

Security checks for vulnerabilities and agentic risk

Overview

This is a transparent Chinese-language decision-making framework with no executable code, though users should verify the package they install and expect local decision notes if they use its memory pattern.

Before installing, confirm the ClawHub package name, publisher, and version resolve to the artifact you intend to use. Be comfortable with Chinese-language instructions, and only allow decision-memory files if you want the agent to keep local records of decisions and outcomes.

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:75
Finding

Unpinned Skill Installation from a Mutable Package Registry

Content
View full analysis

Vulnerability Details

File Location: SKILL.md:75-78 and README.md:8-11
Vulnerability Type: T08: Insecure Dependencies
Risk Level: Medium

Vulnerable Code

SKILL.md:75-78:

bash
## Installation Command
clawhub install decision-framework

README.md:8-11:

bash
## Quick Installation
clawhub install decision-framework

Technical Analysis

The documented installation command identifies the skill only by package name. It does not pin an immutable version, content digest, source revision, or verified artifact checksum. The documentation also does not require verification of a publisher signature or the contents resolved by the package registry.

Consequently, the installed content can differ from the content reviewed in this audit if the registry entry is updated, compromised, transferred to another publisher, or affected by package-name substitution. This is a supply-chain risk: the effective package is resolved at installation time from external mutable state rather than being cryptographically bound to the audited artifact.

The reviewed project itself contains only Markdown documentation and does not include executable scripts. Therefore, this finding does not establish that the currently published dependency is malicious. It identifies an unsafe installation pattern that could permit unaudited future content to be installed.

Attack Path

  1. An attacker compromises the package registry account, registry infrastructure, or package publication process for decision-framework.
  2. The attacker publishes a modified package under the same package name.
  3. A user follows the documented clawhub install decision-framework command.
  4. The package manager resolves and installs the attacker-controlled package because no immutable version or digest is specified.
  5. If the substituted package contains executable hooks, malicious scripts, or hostile agent instructions, those compo ...[truncated 702 chars]
Remediation
View remediation

Remediation Suggestions

  1. Pin the installation to an immutable package version and, where supported, a cryptographic content digest.
  2. Publish and verify a SHA-256 or stronger checksum for the exact package artifact.
  3. Require registry publisher-signature verification before installation.
  4. Document the expected registry and publisher identity explicitly to reduce package-substitution risk.
  5. Inspect the resolved package manifest, scripts, hooks, and skill instructions before activation.
  6. Perform installation and initial execution in a sandbox with minimal filesystem, network, credential, and tool permissions.
  7. Add a lockfile or signed provenance statement binding the documented release to the reviewed source revision.
  8. Update both SKILL.md and README.md so all installation examples use the same verified, immutable release reference.
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • 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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (4)

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The heartbeat instructions are written entirely in Chinese, and the file does not indicate that language selection is optional or that the skill is intended only for Chinese-language use. This creates a natural-language locale constraint without user opt-in, which matches the policy-violation category.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
82% confidence
Finding

This markdown file contains the substantive operating instructions in Chinese, while the skill heading and package identifiers are in English. Because the skill does not state that it is Chinese-only, region-specific, or that users may choose their preferred language, it effectively imposes a language/locale constraint without opt-in.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The protocol section is written entirely in Chinese, including the operative verification steps, with no indication that users may choose another language or that the skill is intentionally limited to a Chinese-language context. This creates a natural-language policy concern because it imposes a specific language without user opt-in.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
88% confidence
Finding

The primary user-facing description is presented only in Chinese ("决策框架Skill") with no indication that other languages are supported or that the skill is intentionally limited to a Chinese-speaking audience. The policy requires avoiding language or locale constraints unless the user is given a choice or the restriction is clearly justified.

Content

No source excerpt is available for this finding.