weekly-report-generator-feishu

• Outcome: Review
• Latest audit: May 29, 2026 · 9:30 PM UTC
• Version: 1.0.2

Overview

This skill does what it says, but it can broadly scan local work history and send the resulting report to Feishu automatically without a clear review step.

Install only if you intentionally want local Git activity and any provided screenshots summarized for Feishu. Before use, set PROJECT_ROOT to a narrow folder, review the generated Markdown before sending, move Feishu secrets out of the script, verify the recipient, and avoid scheduled launchd sending unless unattended recurring reports are acceptable.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands

Findings (15)

Intent-Code Divergence

Medium

• Confidence: 88% confidence
• Finding: The skill contradicts itself by both requiring confirmation and requiring immediate execution/auto-send without asking. In practice, such ambiguity is dangerous because an agent may follow the more permissive path and collect or transmit work data before the user has approved the scope or destination.

Context-Inappropriate Capability

High

• Confidence: 98% confidence
• Finding: The workflow mandates scanning all Git repositories under a configured root, rather than the specific repository or set of repositories the user selected. This is overbroad data access for a weekly report request and can expose unrelated projects, confidential commit messages, branch names, and activity metadata.

Vague Triggers

Medium

• Confidence: 84% confidence
• Finding: The hard-gate instruction forces invocation whenever the user asks for work summaries or Git analysis, without enough scope constraints. Overly broad triggering increases the chance that the skill activates in contexts where the user did not intend repository scanning, script execution, or external sharing.

Vague Triggers

Medium

• Confidence: 82% confidence
• Finding: The invocation examples are generic phrases like generating summaries or viewing commit history, which can cause the skill to run in situations broader than report generation. That makes unanticipated local data access more likely, especially when paired with automatic scanning behavior.

Missing User Warnings

High

• Confidence: 98% confidence
• Finding: The skill instructs automatic scanning of local directories and repositories without a clear user-facing warning that local work data will be accessed. Because commit logs and repository metadata often contain sensitive internal project information, silent scanning materially raises confidentiality risk.

Missing User Warnings

High

• Confidence: 99% confidence
• Finding: The skill mandates automatic sending of the generated report to Feishu immediately and without review or confirmation. This creates direct exfiltration risk because internal development activity, project names, bug details, and screenshots could be transmitted to an external platform unintentionally.

Missing User Warnings

Medium

• Confidence: 90% confidence
• Finding: The script instructs users to hardcode Feishu App ID, App Secret, and recipient identifiers directly in the file, which creates a credential exposure risk if the script is stored in source control, shared, or left on disk with weak permissions. It also later exports sensitive values into the environment, increasing accidental disclosure surface to subprocesses and logs.

Missing User Warnings

Medium

• Confidence: 94% confidence
• Finding: The script sends the weekly report content to Feishu's external API, which is an intentional feature, but there is no explicit consent prompt or warning at send time about transmitting potentially sensitive local work data off-host. In this skill's context, the report may summarize code changes, project names, bugs, and internal business details, so silent transmission materially increases data leakage risk.

Vague Triggers

Medium

• Confidence: 92% confidence
• Finding: The documented trigger phrase is extremely broad and overlaps with ordinary user language, increasing the chance that the skill activates when the user is only discussing weekly reports rather than consenting to repository scanning, screenshot analysis, and message sending. In this skill, accidental activation is more dangerous because activation leads to downstream data collection and external transmission to Feishu.

Vague Triggers

Medium

• Confidence: 89% confidence
• Finding: The usage examples present broad natural-language prompts like normal conversation, without defining activation boundaries or exclusion conditions. That makes unintended execution plausible, and in this context unintended execution can expose local development metadata and work artifacts or trigger outbound messaging without clear consent.

Missing User Warnings

High

• Confidence: 97% confidence
• Finding: The document advertises automatic Feishu delivery and scheduled execution but does not clearly warn users that locally derived content will be transmitted to an external service, possibly without a real-time review step. This creates a meaningful risk of confidential project details, internal metrics, or personal information being sent automatically on a timer or after a casual prompt.

Missing User Warnings

High

• Confidence: 96% confidence
• Finding: The skill claims it can scan multiple Git repositories and recognize screenshot contents, but it omits warnings that these sources may contain secrets, private code, client names, credentials in commit history, meeting notes, or personal data embedded in images. The context makes this especially dangerous because the tool is designed to aggregate data across many local projects, amplifying the volume and sensitivity of exposed information.

Missing User Warnings

Medium

• Confidence: 97% confidence
• Finding: The skill explicitly states that the AI will automatically call the Feishu sending script after generating the report, but the documentation does not present this as a clear consent or privacy warning before use. Because weekly reports are built from local git history and possibly screenshots/work descriptions, users may unknowingly cause potentially sensitive internal information to be exfiltrated to an external messaging platform.

Ssd 3

Medium

• Confidence: 97% confidence
• Finding: The skill requires collection of user and workplace data from repositories and possibly screenshots before obtaining confirmation. This is dangerous because it normalizes non-consensual access to potentially sensitive engineering artifacts, which may include proprietary feature names, incident details, and customer references.

Ssd 3

High

• Confidence: 99% confidence
• Finding: The skill explicitly instructs automatic transmission of the report to an external platform without user review or consent. In this context, that is especially dangerous because the report is synthesized from local Git history and screenshots, amplifying the risk of leaking confidential development and business information.

VirusTotal

66/66 vendors flagged this skill as clean.

Static analysis

No suspicious patterns detected.