Back to skill

Security audit

Tongdaxin

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly a coherent read-only Tongdaxin connector, but its setup instructions tell users to execute a remote installer directly in a shell without integrity checks.

Review the installer before running it. Prefer installing the oo CLI from a pinned, signed, or checksum-verifiable release, and only use this skill for read-only Tongdaxin data operations through an OOMOL account you intend to connect.

Vulnerability Patterns
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • 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
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Findings (1)

T03 · Remote Payload Retrieval and Execution

Error
Location
SKILL.md:56
Finding
Unverified Remote Installer Download and Immediate Execution## Vulnerability Details **File Location**: `SKILL.md`, lines 56–60 **Vulnerability Type**: Remote payload retrieval and execution **Risk Level**: Critical ### Vulnerable Code ```bash curl -fsSL https://cli.oomol.com/install.sh | bash # macOS / Linux ``` ```powershell irm https://cli.oomol.com/install.ps1 | iex # Windows PowerShell ``` ### Technical Analysis The setup instructions pipe responses retrieved from external URLs directly into command interpreters. Neither installation path pins a release, verifies a cryptographic checksum or signature, saves the script for inspection, nor constrains the commands that the downloaded script may execute. HTTPS provides transport protection but does not establish the integrity or immutability of the effective payload after this Skill has been reviewed. Compromise of the remote server, domain, CDN, publishing credentials, or installer-generation process could cause arbitrary replacement code to be executed. Installing the required CLI is related to the Skill's setup, but immediate execution of mutable remote content is not the minimum privilege or safest mechanism necessary to install it. It also falls outside the declared `allowed-tools: [Bash(oo *)]` operational boundary because installation invokes `curl`, `bash`, or PowerShell rather than an already installed `oo` command. ### Attack Path 1. A Tongdaxin operation fails because the `oo` CLI is not installed. 2. The user or agent follows the documented first-time setup instructions. 3. `curl` or `Invoke-RestMethod` retrieves the current installer response from the external OOMOL endpoint. 4. The response is passed directly to Bash or PowerShell without inspection or integrity verification. 5. If the hosting or software-distribution channel has been compromised, attacker-controlled commands execute with the privileges of the invoking account. ### Impact Assessment A substituted installer can execute arbi ...[truncated 734 chars]
Remediation
## Remediation Suggestions 1. Remove both direct execution pipelines (`curl | bash` and `irm | iex`). 2. Direct users to a versioned release from a documented official distribution repository. 3. Download the installer or binary to disk without executing it automatically. 4. Pin an explicit version and publish a cryptographic checksum or verifiable digital signature through an independent trusted channel. 5. Require verification of the checksum or signature before installation. 6. Allow the user to inspect the installer and explicitly approve execution. 7. Prefer signed, platform-native packages with constrained installation behavior. 8. Keep installation outside the Skill's normal `oo`-only tool scope and never trigger it automatically after a failed action. 9. Document the files, permissions, and network endpoints used by the installer so users can evaluate its privilege requirements.
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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 (1)

Vague Triggers

Medium
Confidence
93% confidence
Finding
The skill description says to use this skill for ANY Tongdaxin request and instead of calling the API directly, which creates an overly broad activation scope. This can cause the agent to route all Tongdaxin-related tasks through a shell-capable wrapper without sufficient task-level narrowing, increasing the chance of unintended command execution paths, misuse in inappropriate contexts, or bypass of safer direct integrations.

Static analysis

No suspicious patterns detected.