Back to skill

Security audit

Feishu Plugin Conflict Fix

Security checks for vulnerabilities and agentic risk

Overview

This skill is a coherent Feishu troubleshooting guide, but it includes unsafe recovery commands and recommends higher-privilege plugin access without enough safeguards.

Review this skill carefully before installing. Use its diagnostic commands first, avoid the complete reset unless you have verified backups and exact paths, pin and verify any package before using `npx`, and only grant Feishu document or message-history permissions if those capabilities are truly required.

Vulnerability Patterns
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
  • 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
Findings (3)

T08 · Insecure Dependencies

Error
Location
SKILL.md:347
Finding

Unpinned Remote Package Download and Execution

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, line 347
Vulnerability Type: Unpinned third-party package execution
Risk Level: High

Vulnerable Code

bash
npx -y @larksuite/openclaw-lark install

Technical Analysis

The emergency recovery procedure instructs users to execute an npm package through npx -y without specifying an exact version or verifying package integrity. The -y option automatically accepts installation prompts, while the missing version constraint causes npm to resolve a mutable package release at execution time.

Consequently, the code reviewed during this audit is not necessarily the code that will execute when a user follows the instructions. If the upstream package, maintainer account, npm publication process, or dependency chain is compromised, arbitrary package lifecycle or installer code can run under the invoking user's account.

Although the scoped package name is consistent with the plugin discussed by the Skill, the project does not provide a lockfile, integrity hash, signature verification, or other mechanism for authenticating the downloaded artifact.

Attack Path

  1. An attacker compromises the referenced npm package, a maintainer account, or a transitive dependency.
  2. The attacker publishes a malicious version that remains eligible for default npm resolution.
  3. A user follows the documented emergency recovery procedure.
  4. npx -y retrieves the mutable package release and automatically proceeds without an interactive approval step.
  5. Package lifecycle hooks or the installer execute with the user's privileges.
  6. The malicious package can read or modify files, access environment variables and credentials available to the user, initiate network connections, or alter the OpenClaw installation.

Impact Assessment

Successful exploitation permits arbitrary code execution with the privileges of the user running the command. The affected scope can inc ...[truncated 282 chars]

Remediation
View remediation

Remediation Suggestions

  • Pin the dependency to an exact, previously audited version, for example @larksuite/openclaw-lark@X.Y.Z.
  • Publish and verify an expected package integrity hash or cryptographic signature before execution.
  • Avoid npx -y; retain an explicit review and confirmation step.
  • Use a lockfile or a trusted internal package mirror with immutable artifacts.
  • Review package lifecycle scripts and the resolved dependency tree before installation.
  • Execute installation with the minimum required privileges and in a restricted environment where practical.
  • Document the expected publisher, package version, checksum, and verification procedure.

T09 · Insecure Skill Coding Practices

Warning
Location
SKILL.md:341
Finding

Overbroad Recursive Plugin Deletion During Emergency Recovery

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, lines 341-347
Vulnerability Type: Unsafe destructive filesystem operation
Risk Level: Medium

Vulnerable Code

bash
cp ~/.openclaw/openclaw.json ~/.openclaw/openclaw.json.bak
rm -rf ~/.openclaw/plugins/feishu*
npx -y @larksuite/openclaw-lark install

Technical Analysis

The recovery procedure recursively deletes every entry under ~/.openclaw/plugins whose name starts with feishu. The wildcard is broader than a single known plugin identifier, and rm -rf performs deletion recursively without confirmation.

The preceding backup protects only openclaw.json; it does not preserve the plugin directories or associated data targeted by the deletion. The procedure also does not enumerate matched paths, validate that they are expected plugin directories, reject symbolic links or unusual filesystem entries, or provide a rollback mechanism.

This is primarily a destructive reliability and security-hardening flaw. It becomes exploitable when an attacker or another process can create a matching entry in the plugin directory or can induce a user to run the recovery procedure.

Attack Path

  1. One or more legitimate or attacker-created entries exist under ~/.openclaw/plugins with names matching feishu*.
  2. The user follows the documented complete-reset procedure.
  3. The shell expands the wildcard to every matching path.
  4. rm -rf deletes all matched files and directories without displaying the resolved targets or requesting confirmation.
  5. Unrelated matching plugins, local customizations, cached data, or other plugin content are lost.
  6. The JSON backup cannot restore the deleted plugin directory contents.

Impact Assessment

The operation can cause irreversible deletion of all matching plugin directories available to the invoking user. The direct impact is loss of plugin code, state, local modifications, and potentially plugin-associa ...[truncated 237 chars]

Remediation
View remediation

Remediation Suggestions

  • Replace the wildcard with the exact, validated plugin directory name.
  • Resolve and display the complete target path before deletion.
  • Require explicit user confirmation after listing all affected files.
  • Verify that the target is located directly beneath the expected plugin directory and is not a symbolic link.
  • Back up the affected plugin directories and relevant state before deletion.
  • Prefer a supported OpenClaw uninstall command over direct filesystem deletion.
  • Stop the relevant service before modification and validate configuration integrity before restarting it.
  • Provide a tested rollback procedure that restores both configuration and plugin content.

T05 · Unauthorized Access and Privilege Escalation

Warning
Location
SKILL.md:225
Finding

Recommended Plugin Configuration Expands Feishu Data Privileges

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, lines 225-233
Vulnerability Type: Excessive application permissions
Risk Level: Medium

Relevant Source Excerpt

The affected comparison states that the recommended official plugin operates with user identity, has document read/write access, and can read complete message history. It then explicitly recommends the official feishu-openclaw-plugin.

text
Official feishu-openclaw-plugin:
Identity: User identity
Document permissions: Read and write
Message history: Complete read access
Streaming output: Supported

Recommendation: Use the official feishu-openclaw-plugin

Technical Analysis

The Skill recommends replacing the built-in plugin with a plugin that is expressly described as having broader privileges: user identity, document write access, and complete message-history access. Resolving a tool-name collision does not inherently require document modification privileges or full historical-message access.

The recommendation does not instruct users to restrict OAuth scopes, disable unnecessary capabilities, separate service identities, or obtain informed approval before granting access to historical communications. This breaks the principle of least privilege when the user's actual task only requires conflict resolution or message delivery.

The broader permissions are not evidence that the plugin is malicious. The risk arises because unnecessary authority increases the impact of a compromised agent, plugin, dependency, or tool invocation.

Attack Path

  1. A user encounters a feishu_chat tool-name conflict.
  2. The user follows the Skill's recommendation and enables the official plugin.
  3. The plugin is authorized using a user identity with document read/write and complete message-history capabilities.
  4. A compromised agent session, plugin component, dependency, or improperly authorized tool call invokes those capabilities.
  5. The c ...[truncated 673 chars]
Remediation
View remediation

Remediation Suggestions

  • Recommend the least-privileged plugin capable of performing the required task.
  • Document the exact Feishu OAuth scopes requested by each plugin.
  • Require explicit user approval before enabling message-history or document-write access.
  • Disable document-write and history-reading capabilities unless a concrete workflow requires them.
  • Use a dedicated Bot or service identity instead of a broad user identity where supported.
  • Apply tenant-level access controls and restrict the plugin to necessary chats, users, and documents.
  • Add guidance for periodically reviewing and revoking unused permissions.
  • Present privilege differences as security trade-offs rather than recommending the broader plugin unconditionally.
Vulnerability Patterns
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (11)

Missing User Warnings

High
Category
Not specified by scanner
Confidence
94% confidence
Finding

The reset script deletes Feishu-related configuration keys and overwrites the primary configuration file without an explicit upfront warning about impact, partial-failure handling, or recovery expectations. Although a backup is created, the guide does not instruct the user how to restore it if jq or file moves fail, which can leave the system misconfigured.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
95% confidence
Finding

The '紧急恢复' section includes destructive actions such as deleting plugin directories and reinstalling components, but it does not prominently warn about service disruption, configuration loss, or the need to verify paths and backups before proceeding. In a troubleshooting context, such commands are likely to be copied verbatim, making accidental damage more likely.

Content

No source excerpt is available for this finding.

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
93% confidence
Finding

The command rm -rf ~/.openclaw/plugins/feishu* performs forced recursive deletion using a wildcard in a user configuration/plugins path. Even though it is narrower than deleting the whole home directory, it is still dangerous in a copy-paste troubleshooting guide because path expansion mistakes, unexpected matches, or directory layout changes can remove more than intended and break the environment.

Content

Scanner excerpt · SKILL.md (reported line 344)May include surrounding context.

md
cp ~/.openclaw/openclaw.json ~/.openclaw/openclaw.json.bak

# 2. 删除配置
rm -rf ~/.openclaw/plugins/feishu*

# 3. 重新安装官方插件
npx -y @larksuite/openclaw-lark install

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
93% confidence
Finding

The command rm -rf ~/.openclaw/plugins/feishu* performs forced recursive deletion using a wildcard in a user configuration/plugins path. Even though it is narrower than deleting the whole home directory, it is still dangerous in a copy-paste troubleshooting guide because path expansion mistakes, unexpected matches, or directory layout changes can remove more than intended and break the environment.

Content

Scanner excerpt · SKILL.md (reported line 344)May include surrounding context.

md
cp ~/.openclaw/openclaw.json ~/.openclaw/openclaw.json.bak

# 2. 删除配置
rm -rf ~/.openclaw/plugins/feishu*

# 3. 重新安装官方插件
npx -y @larksuite/openclaw-lark install

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The file consistently presents all operational guidance in Chinese and does not indicate that the skill is Chinese-only or offer an alternative language. This can violate language/locale policy when users have not opted into that language or when the locale limitation is undocumented.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The trigger conditions include broad phrases like '飞书冲突', '工具冲突', and 'TTS 语音', which may cause the skill to activate in situations that do not actually require configuration changes. Over-broad invocation raises the chance that users are shown risky restart, disablement, or reset instructions unnecessarily.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
84% confidence
Finding

The conflict-fix skill includes plugin reinstallation steps that go beyond narrowly scoped diagnosis and configuration repair. Expanding the procedure into package installation increases attack surface and can normalize executing remote code as part of routine troubleshooting, which is risky in an operations guide.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
88% confidence
Finding

The skill instructs users to run npx -y @larksuite/openclaw-lark install without pinning a specific package version. This creates a supply-chain risk because a future compromised or incompatible release could be fetched and executed during a recovery flow, especially in a guide that presents the step as an authoritative fix.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
86% confidence
Finding

The manifest description is extremely brief and does not clearly constrain when the skill should be invoked or what actions it is allowed to perform. In an agent ecosystem, vague trigger scope can cause over-broad activation or misuse of the skill in unrelated contexts, increasing the chance of unintended tool execution or conflict-handling logic being applied where it should not be.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Low
Category
Not specified by scanner
Confidence
80% confidence
Finding

The skill recommends the official plugin partly because it offers broader capabilities such as user-identity access, document write access, and full message-history reading, which exceed the stated goal of resolving tool conflicts. Promoting a more privileged plugin in a conflict-fix guide can steer operators toward unnecessary permissions and enlarge the blast radius if the plugin is misused or compromised.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
78% confidence
Finding

The description is written only in Chinese and provides no indication that other languages are supported or that the language choice is intentional and user-selected. This can be a natural-language policy concern when a skill implicitly enforces a locale or language without opt-in or documented regional justification.

Content

No source excerpt is available for this finding.