ClawHub
Back to skill

Security audit

see-dance-2-video-prompt-architect

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly a coherent video-prompt helper, but its optional setup tells users to execute an unpinned GitHub MCP package locally.

Use the text-only skill normally. Before installing the optional MCP, review the GitHub source yourself, prefer a pinned commit or signed release if available, and understand that `npx -y` will run remote code with your local user permissions and persist an MCP server entry in your agent configuration.

A.I.G

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

Warning
Location
SKILL.md:69
Finding

Unpinned Remote MCP Package Is Downloaded and Executed

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, lines 69–80
Vulnerability Type: Unpinned remote payload retrieval and execution
Risk Level: Medium

Vulnerable Code

bash
Connect the stdio server with:

openclaw mcp add see-dance-2 \
  --command npx \
  --arg -y \
  --arg github:gpt-img-2/see-dance-2-prompt-mcp \
  --include 'build_seedance_prompt,plan_reference_sequence,diagnose_seedance_prompt,get_see_dance_2_resources'

The fallback instruction has the same issue:

text
Then run `openclaw mcp doctor see-dance-2 --probe`. If the installed release lacks the `mcp` command group, add `npx -y github:gpt-img-2/see-dance-2-prompt-mcp` as a Stdio server in MCP settings.

Technical Analysis

The Skill instructs users to execute a package directly from a mutable GitHub repository through npx. The reference does not specify an immutable commit hash, versioned release, integrity checksum, or cryptographic signature. Consequently, the code executed by future installations may differ from the code that existed when the Skill was reviewed.

The -y option automatically accepts the installation prompt, reducing the opportunity for users to inspect or reject the downloaded package. Although the Skill describes the MCP as deterministic and read-only, the MCP implementation is not included in the audited project. Its behavior and claimed restrictions therefore cannot be verified from this artifact.

This most closely matches T03: Remote Payload Retrieval and Execution because the documented procedure retrieves executable code from an external, mutable source and runs it locally.

Attack Path

  1. An attacker compromises the referenced GitHub account, repository, dependency chain, or package installation process.
  2. The attacker modifies the repository so its package lifecycle logic or MCP server entry point executes malicious code.
  3. A user follows the installation instructions in SKILL.md.
  4. `npx -y github:gpt-img-2/see-dance ...[truncated 1084 chars]
Remediation
View remediation

Remediation Suggestions

  1. Pin the GitHub dependency to a reviewed immutable commit hash rather than the repository’s mutable default branch.
  2. Prefer a versioned, signed release from a trusted package registry with a lockfile and integrity metadata.
  3. Publish and verify a cryptographic checksum or signature before execution.
  4. Remove -y so users retain an explicit confirmation step and can review the source being installed.
  5. Vendor the MCP source into the audited project when practical, allowing the executable code and Skill instructions to be reviewed together.
  6. Run the MCP with least privilege in a sandbox or container with restricted filesystem, environment-variable, credential, and network access.
  7. Document the exact permissions and network destinations required by the MCP.
  8. Add automated dependency and source-integrity checks to detect unexpected upstream changes before installation.

SkillSpector

Vulnerability Patterns
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (2)

Rp1

Medium
Category
MCP Rug Pull
Confidence
95% confidence
Finding

The skill instructs users to run an unpinned GitHub package via npx -y github:gpt-img-2/see-dance-2-prompt-mcp, which fetches and executes remote code at install/runtime without a fixed commit, tag, checksum, or signed release. If the upstream repository is compromised or updated maliciously, users following the skill could execute attacker-controlled code on their local machine.

Content

No source excerpt is available for this finding.

Skill Enumeration

Medium
Category
Agent Snooping
Confidence
80% confidence
Finding

Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.

Content

Scanner excerpt · SKILL.md (reported line 94)May include surrounding context.

md
- Seedance 2.0 prompt library: https://see-dance-2.com/prompts/seedance-2-0
- Prompt architecture guide: https://see-dance-2.com/blog/awesome-seedance-2-0-prompts-guide
- Omni reference guide: https://see-dance-2.com/blog/seedance-2-0-omni-reference-guide
- Raw Skill: https://see-dance-2.com/skills/see-dance-2-video-prompt-architect/SKILL.md
- MCP source: https://github.com/gpt-img-2/see-dance-2-prompt-mcp
- ClawHub listing: https://clawhub.ai/gpt-img-2/skills/see-dance-2-video-prompt-architect

Static analysis

No suspicious patterns detected.