Back to skill

Security audit

Workflow Patterns

Security checks for vulnerabilities and agentic risk

Overview

This skill is a straightforward TDD workflow guide; its main risk is that the documented install commands use mutable npx sources, so users should install deliberately.

Before installing, prefer a pinned ClawHub package version or reviewed commit instead of @latest or a mutable GitHub path. Once installed, expect the skill to guide agents through tests, coverage checks, plan edits, and git commits, with explicit approval required at phase checkpoints.

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
README.md:25
Finding
Unpinned Third-Party Code Execution During Installation## Vulnerability Details **File Locations**: - `README.md:25-32` - `SKILL.md:21-23` **Vulnerability Type**: Unpinned and mutable third-party dependencies executed through `npx` **Risk Level**: Medium ### Vulnerable Code `README.md:25-32`: ```bash npx add https://github.com/wpank/ai/tree/main/skills/meta/workflow-patterns ``` ```bash npx clawhub@latest install workflow-patterns ``` `SKILL.md:21-23`: ```bash npx clawhub@latest install workflow-patterns ``` ### Technical Analysis The documented installation commands instruct users to execute third-party content using `npx`. If a requested package is not already available locally, `npx` can retrieve and execute it, including its command-line entry point and potentially package lifecycle scripts. The `clawhub@latest` dependency uses a mutable version selector rather than an audited, exact version. The GitHub-based command similarly references remote repository content without pinning it to a reviewed commit hash or verifying its integrity. Consequently, the code executed when a user follows these instructions can differ from the code available when this project was audited. This does not establish that either current upstream source is malicious. The vulnerability is the lack of controls preventing a compromised registry account, package release, repository, branch, or transitive dependency from supplying altered executable content. ### Attack Path 1. An attacker compromises the package publisher, package registry account, upstream repository, mutable release channel, or a dependency used by the installer. 2. The attacker publishes malicious code under the version resolved by `@latest` or changes the remotely referenced repository content. 3. A user follows one of the documented `npx` installation commands. 4. `npx` downloads the altered package or installer from the external source. 5. The downloaded CLI code or package lif ...[truncated 1052 chars]
Remediation
## Remediation Suggestions 1. Replace `clawhub@latest` with an exact, reviewed package version: ```bash npx clawhub@<reviewed-exact-version> install workflow-patterns ``` 2. Pin GitHub-hosted content to an immutable, reviewed commit hash rather than a mutable branch or repository path. 3. Publish and verify cryptographic checksums or signed release artifacts before installation. 4. Prefer a download, verification, and review workflow over immediately executing remotely resolved content. 5. Use lockfiles and package-manager integrity metadata for all direct and transitive dependencies where supported. 6. Document that installation should be performed without administrative privileges and in an isolated development environment. 7. Add an automated release process that verifies dependency provenance, signatures, integrity hashes, and the exact source revision associated with each documented installation command.
Vulnerability Patterns
  • 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
  • Rogue AgentSelf-Modification, Session Persistence
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (9)

Rp1

Medium
Category
MCP Rug Pull
Confidence
90% confidence
Finding
The README instructs users to execute a remote package/tool via `npx add` from a GitHub URL without any pinned version, tag, or commit digest. This creates a supply-chain risk because future changes to the referenced content could alter what gets installed or executed, and users cannot reproduce or verify the exact artifact they received.

Rp1

Medium
Category
MCP Rug Pull
Confidence
88% confidence
Finding
The command `npx clawhub@latest install workflow-patterns` pulls code at the moving `latest` version rather than an immutable release. If the upstream package is compromised or changes unexpectedly, users may execute different code than intended, making this a classic dependency/supply-chain exposure.

Session Persistence

Medium
Category
Rogue Agent
Content
From your project root:

```bash
mkdir -p .cursor/skills
cp -r ~/.ai-skills/skills/meta/workflow-patterns .cursor/skills/workflow-patterns
```
Confidence
60% confidence
Finding
Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

Skill Enumeration

Medium
Category
Agent Snooping
Content
From your project root:

```bash
mkdir -p .claude/skills
cp -r ~/.ai-skills/skills/meta/workflow-patterns .claude/skills/workflow-patterns
```
Confidence
85% confidence
Finding
Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.

Skill Enumeration

Medium
Category
Agent Snooping
Content
#### Claude Code (global)

```bash
mkdir -p ~/.claude/skills
cp -r ~/.ai-skills/skills/meta/workflow-patterns ~/.claude/skills/workflow-patterns
```
Confidence
85% confidence
Finding
Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.

External Model or Provider Selection

Medium
Category
Excessive Agency
Content
---
name: workflow-patterns
model: standard
version: 2.0.0
description: >
  Systematic task implementation using TDD, phase checkpoints, and structured commits.
Confidence
90% confidence
Finding
Skill selects an external model or provider that may use a different account or billing plan than the operator expects. Undisclosed model switches can cause unexpected cost or quota consumption.

Rp1

Medium
Category
MCP Rug Pull
Confidence
92% confidence
Finding
The installation command uses `npx clawhub@latest`, which fetches and executes the latest published package version at install time. This creates a supply-chain risk because a compromised or malicious newly published version could be executed without review or reproducibility controls.

Autonomous Decision Making

Medium
Category
Excessive Agency
Content
| Making tests pass "too well" | Write minimum code only |
| Skipping refactor | Refactor is where quality happens |
| Ignoring coverage | Add tests until 80%+ |
| Proceeding without approval | Wait for explicit "approved" |

## Task Status Quick Reference
Confidence
75% confidence
Finding
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Autonomous Decision Making

Medium
Category
Excessive Agency
Content
| Making tests pass "too well" | Write minimum code only |
| Skipping refactor | Refactor is where quality happens |
| Ignoring coverage | Add tests until 80%+ |
| Proceeding without approval | Wait for explicit "approved" |

## Task Status Quick Reference
Confidence
75% confidence
Finding
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Static analysis

No suspicious patterns detected.