Back to skill

Security audit

PR Quality Gate

Security checks for vulnerabilities and agentic risk

Overview

This skill is a coherent PR review automation workflow, but it can change GitHub repository state and invoke an external AI CLI without enough scoping or confirmation controls.

Review this skill before installing in repositories where GitHub Issues, PR comments, statuses, or merge actions matter. Use it only with clear user approval for write actions, confirm the target repository and PR, and treat kimi-cli use as potential third-party sharing of code and security findings.

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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (11)

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The README explicitly describes automatic creation of GitHub Issues, PR comments, status updates, and merge recommendations, but does not warn users that invoking the skill can modify repository state. This creates a consent and safety problem: a user may expect analysis-only behavior while the workflow performs write actions against a live repository, potentially generating noise, changing review state, or affecting team processes.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The workflow states that high-severity findings trigger kimi-cli to generate remediation suggestions, implying PR diffs or repository content may be sent to an external AI tool, yet no warning about third-party data sharing is provided. This is dangerous because source code, security issues, or proprietary repository data could be disclosed outside GitHub and the local environment without informed user consent.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The manifest description, category, and trigger list are primarily presented in Chinese alongside limited English terms, and the document does not state that users may choose their preferred language. Under the language/locale policy, a skill should not implicitly enforce a specific language unless it is optional or clearly justified as region-specific.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The trigger phrases are broad and overlap with ordinary review-related conversation, making accidental activation likely. Because this skill can orchestrate issue creation and potentially invoke merge-related GitHub actions, unintended triggering could cause unauthorized repository side effects or surprise users with agent actions they did not explicitly request.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill documentation is internally inconsistent: it claims the skill will not automatically merge PRs, yet elsewhere explicitly includes a final merge step via gh pr merge. In an agentic workflow, this ambiguity is dangerous because it can cause users or downstream orchestrators to permit merge-capable execution under the mistaken belief that merges require separate confirmation.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The manifest states that Medium+ findings will automatically create GitHub Issues, but it does not indicate any user confirmation, dry-run mode, or warning that repository state will be modified. This can cause spam, unwanted disclosure of code/security details into issues, or abuse of repository automation if the workflow is triggered on untrusted input or noisy analysis results.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The workflow describes posting PR comments, updating status, and mentioning developers without warning that it will perform repository-side actions. In practice, this can leak analysis content into public or shared channels, create notification spam, and let an over-broadly triggered workflow affect collaboration or CI decision signals unexpectedly.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
84% confidence
Finding

The workflow accepts a very broad trigger input ('PR URL 或 owner/repo#number') without any stated restrictions on repository scope, trust boundary, or caller authorization. In a skill that reads diffs, creates issues, runs a CLI via PTY, and posts back to GitHub, loose activation constraints increase the chance of operating on unintended or attacker-controlled PRs and amplifying downstream side effects.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
74% confidence
Finding

The README content, examples, and system behavior are entirely presented in Chinese, with no indication that users may opt into another language. This can violate a language/locale policy when a skill constrains output language without explicit user choice or documented justification.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
70% confidence
Finding

Most user-facing descriptions, purposes, and actions are written in Chinese, while the manifest does not indicate that language is optional, configurable, or limited to a China-specific audience. This can violate language/locale policy when a skill implicitly forces one language without user opt-in.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
84% confidence
Finding

The manifest explicitly invokes 'kimi-cli' through a PTY to generate fixes, but provides no warning about subprocess execution or the trust model for the tool and its inputs. PTY-based execution raises the risk of command misuse, prompt-to-command bridging, and unsafe handling of attacker-influenced code snippets when analyzing untrusted PR content.

Content

No source excerpt is available for this finding.