Back to skill

Security audit

office-collaboration-radar

Security checks for vulnerabilities and agentic risk

Overview

ClawScan rates this skill benign with medium confidence: it reports purpose-aligned collaboration-card tools and treats the flagged file-path and Chinese-language behaviors as expected or limited findings.

The verdict is benign, not a guarantee of zero risk. Before installing, consider whether you are comfortable with commands and self-tests using fixed or automatically selected local paths and with some output being required in Chinese. The Skillspector result includes a nonzero exit status and serializer warnings, so its evidence has a processing caveat even though the scan metadata reports complete component coverage.

Vulnerability Patterns
  • 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
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • 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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (12)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

Scoring and visualization are reasonable supporting features for collaboration-status analysis. Normal scores and generate commands require explicit input/output path arguments. However, _selftest() uses tempfile.TemporaryDirectory() and out.write_text() to create a file at a path the user did not explicitly specify, directly contradicting the description's unconditional local-file path restriction. Temporary cleanup does not remove that mismatch. No network access, reminders, or external-system writes are visible in the supplied chunk.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The export functionality is a supporting capability for collaboration status cards; generating Feishu/Notion JSON does not itself write to those external services. The concrete mismatch is in selftest(): tempfile.TemporaryDirectory() selects a local directory without a user-specified path, bad.write_text() writes a file there, and _load_json() is called on generated paths. This contradicts the description's unconditional requirement that all local reads and writes use user-explicit paths. The behavior requires invoking selftest and is not automatic during normal export. Implementations of the imported process helpers are not supplied, so their internal behavior cannot be verified.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding

The main processing functionality matches the declared purpose, and no reminder sending or network access is visible. The mismatch is narrow but explicit: the description requires user-specified paths for all local reads and writes, while _selftest uses tempfile.TemporaryDirectory() and creates, reads, and writes files such as s.txt, bad.json, arr.json, and result.json without user-specified paths. Although these operations support testing, they contradict the description's unconditional file-access restriction.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding

生成雷达图和导出 CSV 可以视为整理协作状态卡片的辅助实现,本身不构成目的偏离。但所示命令直接指定固定输入和输出路径,未体现声明中要求的用户明确指定路径这一限制。代码片段未显示自动触发、发送催办或写入外部系统,因此不据此推断其他不匹配行为。

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
88% confidence
Finding

The enforcement operation appears consistent with preparing traceable status cards. However, the command hardcodes file paths rather than obtaining or verifying user-specified paths, conflicting with the description's explicit local-file-access restriction. The supplied chunk does not establish external writes, reminders, or automatic triggering.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
87% confidence
Finding

卡片聚合本身与声明的协作信息整理用途一致,属于合理的实现细节。差异在于文件访问范围:所示命令使用预设文件名,而声明明确要求所有本地读写路径均由用户指定。若这些路径实际上来自用户的明确指定,则该差异不成立;但当前代码块未提供此依据。没有证据表明存在外部系统写入、催办或额外触发行为。

Content

No source excerpt is available for this finding.

The module mandates Chinese error messages without a user language choice.

Low
Category
Output Handling
Confidence
94% confidence
Finding

The module docstring explicitly requires all invalid inputs to receive friendly Chinese messages. This imposes a specific output language without offering user opt-in or documenting a locale-specific justification.

Content

No source excerpt is available for this finding.

The self-test reads and writes automatically chosen local paths despite the manifest requiring explicitly user-specified paths for all file access.

Low
Category
Excessive Agency
Confidence
97% confidence
Finding

At L270–L274, selftest() creates a temporary directory, writes bad.json, and reads it back; L283 additionally attempts to read another generated path. The selftest command accepts no user-specified paths (L153–L158), so these operations violate the manifest's explicit local-file access restriction, even though they use synthetic test data.

Content

No source excerpt is available for this finding.

CLI error messages are explicitly required to be in Chinese without a user language choice.

Low
Category
Output Handling
Confidence
91% confidence
Finding

The exception documentation specifies that the CLI prints Chinese error messages. This imposes a particular output language without a locale option, user opt-in, or documented region-specific justification.

Content

No source excerpt is available for this finding.

The selftest command reads and writes temporary files at automatically generated paths, contrary to the manifest's explicit-path requirement.

Low
Category
Excessive Agency
Confidence
98% confidence
Finding

The manifest requires all local file reads and writes to use paths explicitly specified by the user. At L0803–L0821, selftest creates an automatically located temporary directory, writes synthetic source and JSON files, and reads them back; the command accepts no user-specified test directory (L1050, L1067–L1068). These operations are confined to temporary test fixtures, but still contradict the stated filesystem boundary.

Content

No source excerpt is available for this finding.

The self-test writes to an automatically selected local path despite the manifest requiring explicitly user-specified paths for all local file access.

Low
Category
Excessive Agency
Confidence
98% confidence
Finding

At L366–L368, the self-test creates a temporary directory and writes test.svg inside it without receiving a user-specified path. Although the temporary directory is cleaned up afterward, this still contradicts the manifest's unconditional restriction on local file reads and writes.

Content

No source excerpt is available for this finding.

Mandates Chinese output labels and placeholder text without user locale opt-in.

Low
Category
Output Handling
Confidence
91% confidence
Finding

The template requires the Chinese heading ## JSON 输出 and the exact Chinese strings 未提供 and 存在冲突,需人工确认 for unknown and conflicting values. These are mandatory natural-language output requirements, with no user language choice or documented locale-specific justification.

Content

No source excerpt is available for this finding.