Back to skill

Security audit

每周安排助理

Security checks for vulnerabilities and agentic risk

Overview

The skill is a coherent local weekly planner, but its command templates could let crafted task text run shell commands if an agent follows them through a shell.

Review before installing. Use it only in an environment where the agent invokes the Python script with structured arguments, not shell-interpolated command strings, and expect it to persist weekly planning data under ./weekly-plans. Consider removing or pinning the unused dependency and tightening triggers before broad use.

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)

T09 · Insecure Skill Coding Practices

Error
Location
SKILL.md:41
Finding

Shell Command Injection Through User-Controlled CLI Arguments

Content
View full analysis
--week --task "" --category "" --date "" --priority "" ``` ```bash python scripts/weekly_planner.py add_completion --year --week --task_id --notes "" ``` ```bash python scripts/weekly_planner.py add_reflection --year --week --date "" --content "" ``` ```bash python scripts/weekly_planner.py update_plan --year --week --task_id --field "" --value "" ``` ```bash python scripts/weekly_planner.py query_by_category --year --week --category "" ``` ### Technical Analysis The Skill instructs the hosting agent to place natural-language values directly inside double-quoted shell command arguments. The affected values include task descriptions, completion notes, reflections, updated field values, and categories. Double quotes do not prevent shell command substitution. If the hosting agent constructs and executes these examples as shell command strings, values containing constructs such as `$(...)` or backticks are interpreted by the shell before `weekly_planner.py` receives them. Quotation marks embedded in input can also terminate the intended argument and introduce additional shell syntax. The Python program uses `argparse` and does not itself invoke a shell. The vulnerability is introduced by the documented execution pattern rather than by `argparse`. Exploitation therefore applies when the agent follows these templates through a shell or another command-string execution interface. Structured process invocation with an argument array would not have this flaw. ## ...[truncated 1332 chars]
Remediation
View remediation

T08 · Insecure Dependencies

Note
Location
SKILL.md:4
Finding

Unbounded and Unnecessary Third-Party Dependency

Content
View full analysis
=2.8.0 system: - mkdir -p ./weekly-plans ``` Dependency description in `SKILL.md`: ```markdown - `python-dateutil>=2.8.0`: date processing and calculation ``` Import in `scripts/weekly_planner.py`: ```python from dateutil import parser ``` ### Technical Analysis The dependency constraint `python-dateutil>=2.8.0` has no upper bound, exact version, or integrity hash. Consequently, separate installations can resolve to different package releases, including releases that were not available or reviewed when the Skill was audited. The imported `dateutil.parser` symbol is not used anywhere in the reviewed Python script. All implemented date calculations use Python's standard `datetime` and `timedelta` functionality. The project therefore introduces package installation and import-time supply-chain exposure without requiring the package for its current functionality. This finding does not establish that `python-dateutil` is malicious. The risk arises from unnecessary third-party code and non-reproducible dependency resolution. ### Attack Path 1. The Skill installation process reads the broad `python-dateutil>=2.8.0` requirement. 2. A package manager resolves a matching version available at installation time. 3. The package and its installation artifacts are downloaded from the configured package source. 4. Installation or import-related package code executes in the Skill environment even though the imported parser is not used. 5. If the configured source, resolved release, package-manager environment, or a future matching release is compromised, malicious code could execute with the installer's or agent's permissi ...[truncated 558 chars]
Remediation
View remediation
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • 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 (11)

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
93% confidence
Finding

The skill performs file reads/writes via local scripts and declares a writable storage path, but it does not define any explicit tool scope such as allowed tools or permissions. In an agent environment, this can let the skill be invoked with broader-than-necessary capabilities, increasing the chance of unintended filesystem access or unsafe execution paths through the referenced script.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The description is broad enough to match common planning, summary, or reporting requests that may fall outside the intended weekly-planner workflow. Over-broad triggering can cause the agent to activate this skill in unrelated contexts, unnecessarily exposing file-writing behavior and creating opportunities for misrouting user data into local persistent storage.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The activation condition says the skill should trigger whenever the user needs to record plans, completions, reflections, generate reports, or view progress, but it lacks constraints on scope, source of truth, and when not to activate. In practice, this ambiguity can lead to over-activation on ordinary chat requests, causing unintended persistence, unexpected script execution, or modification of weekly records without sufficiently specific consent.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The markdown defines a specific trigger phrase but then expands it with '或类似表达' ('or similar expressions'), which leaves the activation boundary unclear. This can overlap with ordinary requests about daily work summaries and makes it ambiguous when the skill should activate versus when it should not.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

This markdown file specifies commands, templates, and operational guidance only in Chinese, including user-facing output templates. Under the policy, forcing a specific language without user opt-in can be a natural-language policy violation unless the locale constraint is explicitly documented and justified, which is not present here.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The entire template is written as a fixed Chinese-language report format, including headings, field labels, and the generation note. This indicates the skill produces output in a specific language by default, with no visible option for users to choose another language or opt in to the locale constraint.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

This Python file contains natural-language descriptions and CLI help text that force a specific language/locale experience for users. Under the policy, language constraints should either be optional for the user or clearly justified as region-specific.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

The docstring says this function adjusts the task date, and it delegates to update_plan(..., "date", new_date). However, task objects are created with the field name assigned_date (L119), and query/report logic reads assigned_date, so this function does not actually change the effective scheduling date used elsewhere.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The argparse descriptions, help messages, warnings, and success messages are all written in Chinese with no opt-in or fallback locale. This creates a language-policy violation because the tool mandates one language rather than letting users choose or documenting a justified regional scope.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
91% confidence
Finding

The file specifies headings, commands context, output labels, and user-facing template text exclusively in Chinese, but does not state that the skill is China-specific or provide a user language option. This can violate language/locale policy where skills should not force a specific language without user opt-in or clear justification.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
95% confidence
Finding

This markdown file contains all user-facing instructions in Chinese and does not indicate that users may choose another language or that the skill is intentionally limited to a Chinese-speaking context. Under the policy, forcing a specific language without user opt-in is a natural-language policy concern.

Content

No source excerpt is available for this finding.