Back to skill

Security audit

huawei-cloud-devbridge-tunnel

Security checks across malware telemetry and agentic risk

Overview

This skill does what it says, but it can silently install software, authenticate to Huawei Cloud, expose local services remotely, run background processes, and delete tunnel resources without enough user control.

Install only if you explicitly want Huawei Cloud DevBridge tunnel automation. Before use, verify the installer source, prefer interactive or temporary credentials, deny anonymous access unless the service is intentionally public, avoid printing tokens in logs or chat, review every exposed port, stop background host processes after use, and require manual confirmation before any delete or delete-all operation.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Rogue AgentSelf-Modification, Session Persistence
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Findings (24)

Lp3

Medium
Category
MCP Least Privilege
Confidence
93% confidence
Finding
The skill clearly invokes shell commands (`curl | bash`, `source`, `nohup`, process checks) yet declares no permissions/capabilities. That mismatch prevents proper policy gating and informed user consent for code execution that modifies the local environment and network exposure.

Context-Inappropriate Capability

Medium
Confidence
90% confidence
Finding
The workflow automatically starts `python3 -m http.server`, which creates and exposes a local service rather than merely managing an existing tunnel. This broadens the skill from tunnel orchestration into arbitrary service hosting, increasing risk of unintended data exposure and side effects on the user's machine.

Vague Triggers

Medium
Confidence
87% confidence
Finding
The trigger phrases include broad terms like `tunnel`, `dev tunnel`, and `host local service`, which can match unrelated requests and cause this high-risk skill to activate unexpectedly. In context, the skill performs installation, authentication, remote exposure, and destructive actions, so over-broad invocation materially raises the chance of unintended execution.

Missing User Warnings

High
Confidence
97% confidence
Finding
The skill instructs the agent to silently install software, change PATH, source scripts, check authentication, and log in automatically before warning the user. In this skill's context, those actions modify the host and can lead directly to exposing local services to remote access without informed consent.

Missing User Warnings

High
Confidence
98% confidence
Finding
The quick-create flow chains installation, authentication, tunnel creation, port creation with anonymous access, and background hosting in one command with minimal visibility. This can unintentionally publish a local service to the internet-like relay surface and leave persistent processes running, creating substantial risk of accidental exposure.

Missing User Warnings

High
Confidence
98% confidence
Finding
Executing destructive operations like tunnel deletion and port deletion without confirmation can destroy user state unexpectedly and disrupt active development workflows. Because this skill also recommends persistent and reusable tunnels, silent deletion can cause broad operational impact across the workspace.

Missing User Warnings

High
Confidence
98% confidence
Finding
The document specifically directs `delete-all` to run without asking the user, despite it being a workspace-wide destructive action. In context, this is especially dangerous because it can remove every active tunnel and break access paths for multiple services or collaborators.

Missing User Warnings

Medium
Confidence
93% confidence
Finding
The document instructs users to run tunnel deletion commands as part of testing but does not warn that these actions permanently remove cloud resources. In an infrastructure-management skill, omission of a deletion warning increases the chance of accidental resource loss, especially if users reuse real tunnel IDs or run steps in production-like environments.

Missing User Warnings

Medium
Confidence
96% confidence
Finding
The token issuance steps direct users to generate host/connect tokens without warning that these are sensitive bearer credentials that can grant tunnel access if exposed. In the context of a tunneling skill, leaked tokens can enable unauthorized hosting or remote connection to exposed local services, making this more dangerous than a generic CLI example.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The hosting and connect procedures instruct users to expose a local service through a tunnel and access it remotely, but they do not warn about the security implications of publishing local services beyond the local machine. Because this skill is specifically designed to expose local ports to remote devices, missing exposure warnings can lead users to unintentionally publish sensitive development services, admin panels, or unauthenticated endpoints.

Missing User Warnings

Medium
Confidence
96% confidence
Finding
The documentation explicitly shows `--access-key` and `--secret-key` being passed on the command line, which can expose credentials through shell history, process listings, audit logs, and terminal recordings. In a skill focused on cloud tunnel management, this is especially risky because compromised IAM credentials could allow unauthorized tunnel creation, token issuance, or broader cloud account access depending on permissions.

Missing User Warnings

Medium
Confidence
94% confidence
Finding
The documentation presents `--anonymous` as a normal option without warning that it can expose a local development service to unauthenticated public access. In the context of a tunneling skill whose purpose is to expose local ports remotely, this materially increases danger because users may unintentionally publish internal apps, admin panels, or debug endpoints to the internet.

Missing User Warnings

Medium
Confidence
98% confidence
Finding
The guide instructs users to fetch and immediately execute a remote shell script with `curl ... | bash`, and states that it will modify shell startup files. This is dangerous because any compromise of the download host, CDN path, or transport chain results in arbitrary code execution and persistent environment changes on the user's machine with no verification step.

Missing User Warnings

Medium
Confidence
99% confidence
Finding
The Windows instructions use `irm ... | iex`, which downloads and executes remote PowerShell code in one step. This creates the same arbitrary code execution risk as pipe-to-shell installs, and on Windows may also run with elevated privileges depending on the user's session.

Missing User Warnings

Medium
Confidence
91% confidence
Finding
The uninstall instructions include `rm -rf ~/.huawei/devbridge`, which removes configuration and authentication material, but the warning is easy to miss and not prominent relative to the destructive command. This can cause credential loss, broken workflows, or accidental deletion of state users intended to preserve.

Missing User Warnings

Medium
Confidence
90% confidence
Finding
The troubleshooting guide recommends destructive actions such as deleting tunnel ports and killing processes, but does not warn users about service disruption, data loss, or the need to verify the target before acting. In an operational context, this can lead to accidental outage of active tunnels or termination of unrelated processes if users follow commands mechanically.

Missing User Warnings

Medium
Confidence
92% confidence
Finding
The verification steps instruct users to host a local HTTP service through a cloud tunnel and connect to it from another device, but they do not warn that this exposes a local service beyond the developer machine. Users may unintentionally publish sensitive local content or development endpoints to remote parties or broader network scopes, especially if they assume this is a purely local test.

Ssd 3

Medium
Confidence
94% confidence
Finding
The skill's normal output format includes printing tunnel tokens in plain text. These tokens function like bearer credentials for host/connect access and may be captured in chat history, logs, screenshots, or terminal recording systems, enabling unauthorized tunnel use.

Autonomous Decision Making

Medium
Category
Excessive Agency
Content
| `--protocol` | `db_port_create` | `http` | `http`, `https`, `auto` | Protocol. |
| `--anon` | `db_port_create` | `allow` | `allow` or `deny` | Anonymous access. |

> **Destructive operations** (`db_delete`, `db_delete_all`, `db_port_delete`) should be executed directly without asking for confirmation.

## Output Format
Confidence
95% confidence
Finding
without asking

Autonomous Decision Making

Medium
Category
Excessive Agency
Content
- **Never hardcode AK/SK** in scripts, logs, or config files. Use `devbridge auth login` or environment variables.
- **Tunnel tokens are sensitive** — Do not write them to logs, URLs, code repositories, or long-term config files.
- **Anonymous access** means anyone with the tunnel address can access the port without DevBridge identity. Enable only for explicitly public content.
- **`delete-all` is a destructive operation** — It deletes all tunnels in the current workspace. Execute directly without asking the user.

### Limitations
Confidence
96% confidence
Finding
without asking

Autonomous Decision Making

Medium
Category
Excessive Agency
Content
## Workflow

> **CRITICAL — Zero confirmation during tunnel creation.** Execute the entire flow (check CLI → load adaptation layer → check auth → create tunnel → configure port → start hosting) **fully automatically with ZERO user interaction**. Do NOT ask the user to confirm parameters or pause between steps. Only surface output after everything is done. **The ONLY exception is tunnel name conflict** — if `db_create` fails because the name exists, present a yes/no choice (是否复用现有隧道?) before proceeding.

### Default Parameters
Confidence
93% confidence
Finding
Do NOT ask the user

Session Persistence

Medium
Category
Rogue Agent
Content
1. **Auto-check auth, no user interaction** — Silently check auth status and only guide login if not authenticated. AK/SK is recommended for automation.
2. **Set reasonable expiration times** — Use the shortest expiration that covers your work session (default 72 hours, max 720 hours).
3. **Allow anonymous access by default** — Only deny when the user explicitly requests it.
4. **Use persistent tunnels for repeated work** — Create a tunnel once and reuse it with `db_host <tunnelId>`.
5. **Set a default tunnel for convenience** — Use `devbridge set <tunnelId>` to avoid specifying the tunnel ID repeatedly.
6. **Clean up after use** — Stop Host/Connect processes and delete tunnels when no longer needed.
7. **Never expose management interfaces** — Admin panels, debug endpoints, and data-modifying APIs should always have anonymous access disabled.
Confidence
82% confidence
Finding
Create a tunnel once and reuse it with `db_host <tunnelId>`. 5. **Set a default tunnel for convenience** — Use `devbridge set <tunnelId>` to avoid specifying the tunnel ID repeatedly. 6. **Clean up af

External Script Fetching

High
Category
Supply Chain
Content
- Verify network connectivity to the installation source.
- Check if a proxy is needed: `echo $http_proxy`
- Retry with an explicit proxy: `curl --proxy <proxy-url> -fsSL <install-url> | bash`
Confidence
99% confidence
Finding
curl --proxy <proxy-url> -fsSL <install-url> | bash

External Script Fetching

High
Category
Supply Chain
Content
db_init() {
    if ! command -v devbridge &>/dev/null; then
        echo "ERROR: devbridge CLI not found. Please install it first."
        echo "Run: curl -fsSL https://res-hd.hc-cdn.cn/sharedata/hdspace/devbridge/install.sh | bash"
        return 1
    fi
    return 0
Confidence
98% confidence
Finding
curl -fsSL https://res-hd.hc-cdn.cn/sharedata/hdspace/devbridge/install.sh | bash

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.