Back to skill

Security audit

瑜伽体式知识库 (CUC)

Security checks for vulnerabilities and agentic risk

Overview

The skill is a coherent yoga-pose helper, but it ships and uses a plaintext external API token with unclear privileges.

Review before installing. The skill appears intended to answer yoga course questions, not to damage local systems, but it contacts an external NocoDB service using a shared token embedded in the package. The publisher should rotate that token, remove it from the artifact, require a user-provided or restricted read-only credential, and add basic yoga safety guidance.

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
scripts/query.sh:5
Finding

Hardcoded NocoDB API Token

Content
View full analysis

Vulnerability Details

File Location: scripts/query.sh, line 5
Vulnerability Type: Hardcoded external-service credential
Risk Level: Medium

Vulnerable Code

bash
XC_TOKEN="${NOCODB_XC_TOKEN:-cCCOb9nNdeWSFIWr7tg7rWfOZeKyTfJ0Qtis_bm3}"

Technical Analysis

The script embeds a NocoDB API token as the default value used whenever the NOCODB_XC_TOKEN environment variable is not defined. Because the token is distributed with the Skill package in plaintext, any person with access to the package can recover and reuse it independently of the Skill.

The script sends this credential in the xc-token header to https://nocodb.dixchain.com/api/v2. The repository does not establish the token's server-side privileges, expiration, or resource restrictions. Consequently, its exact authorization scope cannot be confirmed through static analysis.

Attack Path

  1. Obtain a copy of the Skill package or source repository.
  2. Read scripts/query.sh and extract the plaintext token from line 5.
  3. Construct requests to the NocoDB API at https://nocodb.dixchain.com/api/v2.
  4. Supply the recovered value in the xc-token request header.
  5. Access, enumerate, or modify resources to the extent permitted by the token's server-side privileges.

No additional local compromise or code execution is required to recover the credential.

Impact Assessment

An attacker can obtain the same NocoDB privileges assigned to the exposed token. Potential consequences include unauthorized access to yoga-pose records, dataset enumeration, API quota consumption, and record modification if write privileges are enabled.

The exposure also prevents reliable attribution because all installations and unauthorized users may operate under the same credential. The maximum impact is limited by the token's actual server-side permissions, which could not be determined from the audited files.

Remediation
View remediation

Remediation Suggestions

  1. Immediately revoke and rotate the exposed token.

  2. Remove the fallback credential from scripts/query.sh.

  3. Require the token to be supplied through an environment variable or secret manager and terminate safely when it is absent:

    bash
    : "${NOCODB_XC_TOKEN:?NOCODB_XC_TOKEN must be configured}"
    XC_TOKEN="$NOCODB_XC_TOKEN"
    
  4. Provision a dedicated read-only token restricted to the required table, view, and API operation.

  5. Apply expiration, rotation, rate limiting, and audit logging to the replacement credential.

  6. Review API logs for unauthorized use of the exposed token.

  7. Purge the credential from repository history and release artifacts where practical.

  8. Add automated secret scanning to the development and release pipeline.

Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (7)

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
92% confidence
Finding

The skill instructs the agent to execute a shell script (bash {skillDir}/scripts/query.sh) but does not declare any explicit tool scope or allowed-tools boundary. This creates an unnecessary capability gap where shell execution may occur without least-privilege constraints, increasing risk if the script or its environment is modified, or if future changes introduce unsafe command construction.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

This skill gives yoga pose guidance, including potentially strenuous or injury-prone postures like backbends and balance poses, but omits safety warnings, contraindications, and advice to stop if pain occurs. Users may rely on the instructions as actionable physical guidance and could be harmed if they have injuries, mobility limitations, or perform the pose incorrectly without cautionary context.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The file gives pose recommendations for goals like relieving back pain, improving posture, and stretching specific body areas without any safety disclaimer, contraindications, or advice to stop if pain occurs. In a yoga instruction context, users may rely on these mappings as fitness or quasi-health guidance and attempt poses that are unsuitable for their condition, increasing risk of strain or worsening existing injuries.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
85% confidence
Finding

描述中将技能定位为中文知识库并直接规定提供相关解答,但没有说明是否支持其他语言或允许用户自行选择语言。按自然语言政策,若技能隐含强制特定语言输出而无用户选择,可能构成语言/地区策略问题。

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
77% confidence
Finding

The skill content appears to force a single language/locale presentation without offering user choice or documenting that the skill is intended only for a Chinese-speaking audience. Under the language/locale policy criterion, monolingual instructions can be a policy concern when no opt-in or justified regional scope is provided.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
79% confidence
Finding

The natural-language comments describe the skill solely in Chinese and do not offer any language choice or indicate that the locale restriction is intentional and justified. Under the stated policy, forcing a specific language without user opt-in can be a language or locale policy violation.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
86% confidence
Finding

This shell script sends a request to an external NocoDB endpoint using curl, which is a safety-relevant network operation for a code file. The file contains no confirmation prompt, user-facing notice, or explanatory comment warning that executing the script will contact a remote service and retrieve data.

Content

No source excerpt is available for this finding.