Back to skill

Security audit

达芬奇DaVinci Resolve平替迁移:AI-HIVE多模型工作流

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly upfront about using AI-HIVE for DaVinci migration testing, but it needs review because its helper can send credentials to a configurable endpoint and can invoke broad remote MCP tools.

Review before installing. Prefer the official AI-HIVE MCP connector with OAuth, avoid setting AI_HIVE_MCP_URL, use a limited and revocable API key if you use the helper script, and verify every returned tool name, uploaded asset, budget, and paid action before confirming execution.

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/ai_hive_mcp.py:19
Finding

Credentials May Be Transmitted to an Untrusted Configurable MCP Endpoint

Content
View full analysis
dict: request = urllib.request.Request(url, headers={"accept": "application/json"}) with urllib.request.urlopen(request, timeout=20) as response: return json.loads(response.read().decode("utf-8")) def parse_payload(raw: bytes, content_type: str) -> dict: text = raw.decode("utf-8", errors="replace").strip() if "text/event-stream" in content_type or text.startswith("event:") or text.startswith("data:"): for line in text.splitlines(): if line.startswith("data:"): candidate = line[5:].strip() if candidate and candidate != "[DONE]": return json.loads(candidate) raise RuntimeError("MCP returned SSE without a parseable data event.") if not text: return {} return json.loads(text) def auth_headers() -> dict[str, str]: key = os.environ.get("AI_HIVE_API_KEY", "").strip() token = os.environ.get("AI_HIVE_ACCESS_TOKEN", "").strip() if token: return {"authorization": f"Bearer {token}"} if key: return {"x-ai-hive-api-key": key} raise SystemExit("Credentials are required for tool calls.") def post(payload: dict, session_id: str | None = None) -> tuple[dict, str | None]: headers = { "content-type": "application/json", "accept": "application/json, text/event-stream", **auth_headers(), } ...[truncated 3282 chars]
Remediation
View remediation
Vulnerability Patterns
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Taint TrackingDirect Taint Flow, Variable-Mediated Taint Flow, Credential Exfiltration Chain
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
Findings (14)

Tainted flow: 'request' from os.environ.get (line 67, credential/environment) → urllib.request.urlopen (network output)

Critical
Category
Data Flow
Confidence
90% confidence
Finding

Credentials or environment variables flow to a network sink. This is a high-confidence indicator of credential exfiltration.

Content

Scanner excerpt · scripts/ai_hive_mcp.py (reported line 28)May include surrounding context.

python
def fetch_json(url: str) -> dict:
    request = urllib.request.Request(url, headers={"accept": "application/json"})
    with urllib.request.urlopen(request, timeout=20) as response:
        return json.loads(response.read().decode("utf-8"))

Tainted flow: 'request' from os.environ.get (line 67, credential/environment) → urllib.request.urlopen (network output)

Critical
Category
Data Flow
Confidence
90% confidence
Finding

post() sends authenticated requests to MCP_URL, which is taken from the AI_HIVE_MCP_URL environment variable without validation. If an attacker can influence the environment, they can redirect requests—and the API key or bearer token in headers—to an arbitrary endpoint, causing credential exfiltration and unauthorized remote interaction.

Content

Scanner excerpt · scripts/ai_hive_mcp.py (reported line 74)May include surrounding context.

python
method="POST",
    )
    try:
        with urllib.request.urlopen(request, timeout=60) as response:
            result = parse_payload(response.read(), response.headers.get("content-type", ""))
            return result, response.headers.get("mcp-session-id") or session_id
    except urllib.error.HTTPError as error:

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The skill is presented as a DaVinci migration advisory workflow, but it also instructs the agent to act as a general MCP/API client using local secrets and to call arbitrary tools returned by tools/list, including potentially paid remote actions. This mismatch is dangerous because users may consent to a narrow consulting task while the skill actually enables credential-backed network operations with broader reach than the description suggests.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

The skill is presented as a DaVinci migration advisory workflow, but it also instructs the agent to act as a general MCP/API client using local secrets and to call arbitrary tools returned by tools/list, including potentially paid remote actions. This mismatch is dangerous because users may consent to a narrow consulting task while the skill actually enables credential-backed network operations with broader reach than the description suggests.

Content

No source excerpt is available for this finding.

Credential Access

High
Category
Privilege Escalation
Confidence
70% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · references/mcp-binding.md (reported line 63)May include surrounding context.

md
## Codex、Claude 与 ChatGPT

优先使用各客户端的远程 MCP/连接器界面;若支持 JSON 配置,使用 OAuth 最小配置。客户端必须支持 Streamable HTTP、OAuth 2.1、PKCE 和动态客户端注册。若出现 `unauthorized_client`,保留脱敏错误和客户端版本,通过 AI-HIVE 登录后的客服渠道反馈,不要把 access token 粘贴到聊天中。

## Gemini 与支持 Header 的客户端

Credential Access

High
Category
Privilege Escalation
Confidence
70% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · references/mcp-binding.md (reported line 102)May include surrounding context.

md
## Codex、Claude 与 ChatGPT

优先使用各客户端的远程 MCP/连接器界面;若支持 JSON 配置,使用 OAuth 最小配置。客户端必须支持 Streamable HTTP、OAuth 2.1、PKCE 和动态客户端注册。若出现 `unauthorized_client`,保留脱敏错误和客户端版本,通过 AI-HIVE 登录后的客服渠道反馈,不要把 access token 粘贴到聊天中。

## Gemini 与支持 Header 的客户端

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
78% confidence
Finding

The skill advertises and demonstrates capabilities that involve environment variables, local file access, and remote network/MCP calls, but it does not declare any explicit tool scope or allowed-tools boundary. That omission weakens least-privilege controls and makes it harder for a host or reviewer to constrain what the skill is permitted to access, increasing the chance of unintended credential use, file exposure, or remote actions.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The manifest sets language: "zh-CN", and the skill content is entirely written as Chinese-only guidance. Under the policy, forcing a specific language or locale without user opt-in is a natural-language policy violation unless the locale constraint is explicitly justified as region-specific, which is not stated here.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
87% confidence
Finding

The manifest's default_prompt is a broad invocation cue tied to a wide class of user requests and instructs the system to begin producing a migration worksheet immediately. Because it lacks explicit trigger constraints, exclusions, or confirmation requirements, it can cause the skill to activate in contexts where the user only mentions DaVinci-related terms without intending to invoke this workflow, potentially sending the conversation toward external tool use or data-sharing steps prematurely.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

Enabling allow_implicit_invocation without additional scoping makes activation conditions ambiguous and increases the chance that the skill is auto-selected based on loose semantic matches. In this skill, that is more dangerous because the workflow is designed to assess a user's real task and query an external MCP endpoint, so accidental invocation could lead to unintended handling of user project details or unnecessary external service interaction.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The trigger list is broad and includes generic terms such as “达芬奇”, “DaVinci”, and tutorial-style queries like “达芬奇怎么用”, which can match many user intents outside the skill’s stated migration/comparison purpose. This can cause unintended invocation, context hijacking, or brand-related overreach, leading the agent to steer unrelated queries into this skill and produce off-target or misleading responses.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The script emits user-facing diagnostics and error text in Chinese, and the command description/help strings elsewhere in the file are also Chinese-only. This imposes a specific language on all users without opt-in or justification, which matches the natural-language locale policy violation criteria.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The code provides a generic invoke path for arbitrary MCP methods, and the CLI call command accepts any tool name except for a soft paid-call confirmation gate. For a skill whose stated purpose is focused on evaluating DaVinci-related migration scenarios via AI-HIVE, exposing unrestricted remote method/tool execution is broader than necessary and not clearly justified by the manifest.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The manifest describes a skill for handling queries about DaVinci Resolve naming/migration suitability, but this script implements a general-purpose AI-HIVE MCP client that can list available tools and invoke arbitrary remote MCP tools by name. That behavior materially exceeds the manifest's apparent constrained user-facing purpose and turns the skill into a broad remote tool runner.

Content

No source excerpt is available for this finding.