Back to skill

Security audit

guaikei-kuaishou-public-comment-fetcher

Security checks across malware telemetry and agentic risk

Overview

This is a coherent Kuaishou public-data tool, but it needs review because its instructions can expose the API token and it automatically saves fetched results locally.

Review before installing. Use this only for public Kuaishou data you are allowed to collect, confirm you are comfortable sending requests through Guaikei, and avoid running or sharing the documented echo $GUAIKEI_API_TOKEN command. Treat generated logs as retained data and delete or protect them if they include sensitive research targets or personal identifiers.

Vulnerability Patterns
  • 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
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
Findings (1)

T09 · Insecure Skill Coding Practices

Warning
Location
SKILL.md:271
Finding
API Token Disclosure Through Troubleshooting Command## Vulnerability Details **File Location**: `SKILL.md:271` **Vulnerability Type**: Plaintext sensitive-data exposure **Risk Level**: Medium **Complete Code Snippet**: ```markdown **Q7. Command exits immediately without producing data?** > Check: in most cases, `GUAIKEI_API_TOKEN` did not pass validation (see Q1). Before running, execute `echo $GUAIKEI_API_TOKEN` to confirm that the variable was injected. ``` ### Technical Analysis The troubleshooting instructions tell users to print the complete `GUAIKEI_API_TOKEN` value to the terminal. Although the application code otherwise sends this credential through an HTTPS header to the fixed API host, explicitly rendering the secret creates an unnecessary disclosure path. Terminal output may be retained in CI/CD logs, remote shell recordings, support transcripts, screen-sharing sessions, or copied diagnostic output. An attacker does not need code execution or elevated local privileges if they can observe or retrieve one of these output channels. ### Attack Path 1. A user encounters a token-validation or startup error. 2. The user follows `SKILL.md:271` and runs `echo $GUAIKEI_API_TOKEN`. 3. The complete API token is displayed in plaintext. 4. The output is observed, recorded, copied into a support request, or retained by an automated logging system. 5. An unauthorized party retrieves the exposed token. 6. The party reuses it to authenticate requests to the Guaikei API until the credential expires or is revoked. ### Impact Assessment Exposure can permit unauthorized use of the affected user's Guaikei API authorization, including consumption of account quotas and access to API operations allowed by that token. The available privileges are bounded by the token's server-side permissions; the reviewed project provides no evidence that the token grants operating-system access or broader local privileges. The issue does not expose unrelated environment variables automaticall ...[truncated 65 chars]
Remediation
## Remediation Suggestions - Remove the instruction to print the token. - Verify only whether the environment variable is present: ```sh if [ -n "${GUAIKEI_API_TOKEN:-}" ]; then echo "GUAIKEI_API_TOKEN is configured" else echo "GUAIKEI_API_TOKEN is missing" fi ``` - If identification is essential, show only a short masked suffix and never the complete value. - Add explicit documentation warning users not to include credentials in logs, screenshots, issue reports, or support transcripts. - Recommend immediate token revocation and rotation if the value has already been exposed. - Ensure CI/CD systems mask `GUAIKEI_API_TOKEN` and related secret values in captured output.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • 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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (6)

Description-Behavior Mismatch

Medium
Confidence
94% confidence
Finding
The manifest description omits that operation requires sending user-supplied targets and retrieved data to a third-party service (`guaikei.com` API). This matters because users and reviewers may not realize that requests, identifiers, and potentially collected public-comment data leave the local environment, which creates data handling, privacy, and compliance risk.

Description-Behavior Mismatch

Low
Confidence
94% confidence
Finding
The manifest description omits that operation requires sending user-supplied targets and retrieved data to a third-party service (`guaikei.com` API). This matters because users and reviewers may not realize that requests, identifiers, and potentially collected public-comment data leave the local environment, which creates data handling, privacy, and compliance risk.

Description-Behavior Mismatch

Medium
Confidence
85% confidence
Finding
The README markets broader Kuaishou data-mining functions such as competitor monitoring, trend prediction, KOL screening, keyword-based discovery, and large-scale public comment collection that go beyond the stated narrower scope of work details/data analysis. This scope expansion is dangerous because users may run or trust capabilities the manifest did not clearly declare, weakening informed consent, review accuracy, and policy enforcement for scraping or surveillance-like behavior.

Description-Behavior Mismatch

Medium
Confidence
89% confidence
Finding
The CLI persists fetched comment data to a local JSON file after completing the request, but the skill description only advertises analytics/detail retrieval and does not indicate local retention. Comments can contain personal data, account identifiers, or sensitive business research artifacts, so silent persistence increases data exposure through leftover files, backups, or later access by other users/processes.

Description-Behavior Mismatch

Low
Confidence
84% confidence
Finding
The command writes fetched results to a local JSON file after successful execution, creating an undisclosed persistence side effect. Even if the data is not highly sensitive by default, silent local storage can expose scraped results, user-supplied targets, and metadata to other local users, backup systems, or downstream tooling that assumes the skill is read-only.

Description-Behavior Mismatch

Medium
Confidence
94% confidence
Finding
The CLI performs keyword-based content search and creates search tasks, which exceeds the declared scope of a skill described as work details/analytics fetching. This scope mismatch can mislead users and reviewers about the data-access and collection behavior of the skill, enabling unintended scraping or broader content discovery than expected.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

Detected: suspicious.exposed_secret_literal

File appears to expose a hardcoded API secret or token.

Critical
Code
suspicious.exposed_secret_literal
Location
SKILL.md:15