Back to skill

Security audit

Daniels Running Plan

Security checks for vulnerabilities and agentic risk

Overview

This skill coherently generates Chinese-language running plans and exports them, but users should know it saves plans to fixed desktop filenames.

Install only if you are comfortable with a Chinese-language skill that may save your age, race goals, schedule, and training plan into TXT and CSV files on the desktop. Check or change the output path before use if the machine is shared, and avoid overwriting existing running-plan files.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • 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
README.md:30
Finding

Unpinned Global npm Package Installation

Content
View full analysis

Vulnerability Details

File Location: README.md, lines 30-35
Vulnerability Type: Unpinned third-party dependency installed globally
Risk Level: Medium

Vulnerable Code Snippet:

bash
npm i -g clawhub
clawhub install daniels-running-plan

Technical Analysis

The installation instructions retrieve the latest available version of the clawhub npm package without specifying a reviewed version or verifying package integrity. npm packages may execute lifecycle scripts during installation. The -g option installs the package globally, increasing its reach beyond the project directory.

The command does not independently prove malicious behavior. However, it creates a supply-chain exposure because the effective code installed in the future can differ from the code reviewed when these instructions were published. Compromise of the package, its publisher account, the configured npm registry, or a future package release could cause users to install attacker-controlled code.

Attack Path

  1. An attacker compromises the clawhub package, its publisher account, its dependency chain, or the package source used by the configured registry.
  2. The attacker publishes a malicious release or modifies a transitive dependency.
  3. A user follows the documented npm i -g clawhub instruction.
  4. npm resolves the mutable latest version and downloads the compromised package.
  5. Malicious lifecycle scripts or installed executable code run with the permissions of the user performing the installation.
  6. The globally installed executable remains available to subsequent sessions until removed.

Impact Assessment

Successful exploitation could execute arbitrary code with the installing user's privileges. The attacker could access or modify files available to that account, steal environment variables or credentials readable by it, alter globally installed tooling, and affect later invocations of the installed comm ...[truncated 202 chars]

Remediation
View remediation

Remediation Suggestions

  • Pin clawhub to an explicitly reviewed version rather than relying on the latest release.
  • Document the expected npm registry and exact package identity.
  • Publish and verify an integrity hash, lockfile, signed provenance, or equivalent package attestation.
  • Prefer a project-scoped installation or an ephemeral invocation over a global installation where feasible.
  • Review package lifecycle scripts and the dependency tree before recommending installation.
  • Avoid recommending elevated installation privileges.
  • Establish a process for reviewing and updating the pinned version when security fixes are released.

T09 · Insecure Skill Coding Practices

Note
Location
SKILL.md:84
Finding

Fixed Output Paths Can Overwrite Existing User Files

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, lines 84-88
Vulnerability Type: Unsafe fixed-name file output
Risk Level: Low

Vulnerable Code Snippet:

text
### Step 5: Save Files

Save both copies to the desktop:
1. TXT version: `/home/ai/桌面/running-plan.txt`
2. CSV version: `/home/ai/桌面/running-plan.csv` using UTF-8 BOM (`encoding='utf-8-sig'`)

Technical Analysis

The skill directs the agent to save every generated plan to two fixed filesystem paths. It does not require checking whether either file already exists, obtaining overwrite confirmation, creating a unique filename, or using an atomic creation mode that refuses replacement.

Consequently, repeated skill executions can silently replace earlier plans. The fixed shared filenames also create collisions between different users, tasks, or concurrent invocations if the runtime reuses the same home directory.

This is a file-integrity issue rather than arbitrary file access: the documented paths are static and do not directly accept attacker-controlled path input. The practical exposure is therefore limited to files already located at the specified destinations and writable by the executing account.

Attack Path

  1. A preexisting file is stored at /home/ai/桌面/running-plan.txt or /home/ai/桌面/running-plan.csv.
  2. The user or an automated workflow invokes the skill to generate another plan.
  3. The agent follows the mandatory save instructions without checking whether the target exists.
  4. The save operation truncates and replaces the existing file.
  5. The prior plan or other content at that fixed destination is lost or becomes unavailable.

Impact Assessment

Exploitation or accidental triggering can modify files writable by the agent account at the two fixed destinations. It may cause loss of previous plans, cross-task data collisions, and incorrect downstream processing of replaced CSV data. No privilege escalation, arbitr ...[truncated 93 chars]

Remediation
View remediation

Remediation Suggestions

  • Ask the user to select or approve the output directory.
  • Generate unique filenames using a safe timestamp, plan identifier, or non-conflicting numeric suffix.
  • Check whether each destination exists before writing and request explicit confirmation before replacement.
  • Use exclusive file creation where supported so an existing file cannot be overwritten accidentally.
  • Write to a temporary file in the destination directory and atomically rename it after successful generation.
  • Avoid shared fixed paths in multi-user or concurrent environments.
  • Clearly report the final paths after saving and preserve prior versions when practical.
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (7)

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
86% confidence
Finding

The skill description and usage instructions are written entirely in Chinese, and line L59 explicitly states the CSV is intended to open in Excel with Chinese text preserved. There is no indication that other languages are supported or that the Chinese-only presentation is an intentional, justified region-specific constraint, which can violate language/locale choice policy.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

L003 的触发条件包含“制定跑步计划”“询问如何训练跑步时触发”等宽泛自然语言表述,缺少明确边界或排除条件。这类描述容易与一般性的跑步建议咨询重叠,增加非预期触发的风险。

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The skill is primarily conversational but instructs writing output files to a fixed local desktop path without establishing necessity, consent, or execution safeguards. This creates an unnecessary side effect: local file creation on the host, which can surprise users, leak sensitive training/health data to shared environments, or be abused if similar patterns are generalized to more sensitive content.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

A running-plan skill does not inherently require local filesystem access, so mandatory export to /home/ai/桌面 is unjustified privilege use. Even if the content is benign, unnecessary write capability expands the attack surface and can expose personal fitness goals, age, or schedule information on the local machine.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The skill directs the agent to save files locally without notifying the user that a filesystem write will occur. Hidden side effects violate user expectations and can lead to unauthorized persistence of personal data, especially on shared or monitored systems.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

This markdown file contains user-facing instructional content exclusively in Chinese, and it does not state that the language is optional or limited to a Chinese-speaking context. Under the language/locale policy, forcing a specific language without user opt-in is a natural-language policy concern.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
88% confidence
Finding

L003 及后续全文均以中文规定技能行为,但没有说明这是面向特定中文用户群的限制,也未给出其他语言选项。根据语言/locale 政策,强制单一语言而无用户选择可能构成自然语言策略违规。

Content

No source excerpt is available for this finding.