Back to skill

Security audit

Neta

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly a routing guide, but its install instructions ask users to run unpinned remote skill installation commands that can change local agent behavior.

Treat this as a routing/index skill plus optional install guide. Before running the `npx skills add` commands, verify that you trust the `skills` npm package and the `talesofai/neta-skills` repository, prefer pinned versions or commit hashes, and review each sub-skill because those installed skills may add broader capabilities such as content creation, community interactions, or account-related changes.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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)

T08 · Insecure Dependencies

Warning
Location
SKILL.md:16
Finding
Unpinned Third-Party Package Execution and Remote Skill Installation## Vulnerability Details **File Location**: `SKILL.md`, lines 16–34 **Vulnerability Type**: Unpinned and mutable third-party dependencies **Risk Level**: Medium ```bash # Spaces and worldbuilding npx skills add talesofai/neta-skills/skills/neta-space # Creative content (images/videos/songs/MVs) npx skills add talesofai/neta-skills/skills/neta-creative # Community browsing and interactions npx skills add talesofai/neta-skills/skills/neta-community # Research and content suggestions npx skills add talesofai/neta-skills/skills/neta-suggest # Character creation and management npx skills add talesofai/neta-skills/skills/neta-character # Elementum (visual style/concept) creation and management npx skills add talesofai/neta-skills/skills/neta-elementum ``` ### Technical Analysis The documented commands invoke `npx skills` without pinning the npm package to a reviewed version. Depending on local npm behavior and package availability, `npx` may retrieve and execute package code from the npm registry. The commands also install skill content from the mutable `talesofai/neta-skills` repository without specifying an immutable commit, release, or integrity hash. Consequently, the effective installer and installed skill content can change after this project has been reviewed. A compromise of the npm package, its maintainer account, the remote repository, or an upstream dependency could introduce attacker-controlled code or skill instructions. Remote installation is relevant when a user explicitly requests a sub-skill, but it is not necessary for this skill's primary overview and routing function. Recommending immediate execution of third-party installers therefore exceeds the minimum privileges required merely to classify a task or explain the capability map. The static pre-scan warning concerning sensitive network transmission was also reviewed. `SKILL.md` does not read credentials, environment variables, private keys, or local files, and it does not explicitly in ...[truncated 1717 chars]
Remediation
## Remediation Suggestions 1. Pin the npm package to an explicitly reviewed version, for example by using a version-qualified package reference supported by the installation environment. 2. Pin each remote skill to an immutable commit hash or signed release rather than a mutable repository path. 3. Publish and verify cryptographic checksums or signatures for the installer and installed skill contents. 4. Require explicit user confirmation before downloading or executing third-party content, and display the source, version, and requested destination. 5. Separate routing guidance from installation. The overview skill should select and describe the appropriate sub-skill without automatically requiring package execution. 6. Recommend reviewing the selected sub-skill and its scripts before installation or invocation. 7. Run installation in a sandbox or least-privileged environment with restricted filesystem, credential, and network access. 8. Document the trusted npm scope, repository owner, expected package identity, and a procedure for responding to upstream compromise.
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • 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 (12)

Rp1

Medium
Category
MCP Rug Pull
Confidence
92% confidence
Finding
The skill instructs users to run `npx skills add ...` without pinning an exact package/version, which means the resolved code can change over time and may pull a compromised or maliciously updated dependency. In a skill-installation context, this is meaningfully risky because the command leads directly to fetching and executing remote package logic.

Rp1

Medium
Category
MCP Rug Pull
Confidence
92% confidence
Finding
This installation example relies on unpinned `npx skills`, allowing whatever current package version resolves at runtime to execute. Because the command is presented as an installation path for a sub-skill, a compromised upstream package or typo-squatted resolution could result in arbitrary code execution in the user's environment.

Rp1

Medium
Category
MCP Rug Pull
Confidence
92% confidence
Finding
The documented use of `npx skills` is unpinned and therefore trusts the latest package resolution at the time of execution. Since `npx` executes fetched package code, this creates a supply-chain risk if the package or one of its dependencies is replaced, hijacked, or maliciously updated.

Rp1

Medium
Category
MCP Rug Pull
Confidence
92% confidence
Finding
The example executes remote package logic via `npx skills` without constraining the version, making the installation path non-reproducible and vulnerable to upstream compromise. In documentation intended to be copied verbatim, this can directly propagate unsafe operational practices to end users.

Rp1

Medium
Category
MCP Rug Pull
Confidence
92% confidence
Finding
The instruction uses `npx skills` with no explicit version, so the code executed depends on the registry state at runtime. That opens a realistic supply-chain attack surface for anyone following the skill's install instructions, especially because `npx` may download and run code immediately.

Rp1

Medium
Category
MCP Rug Pull
Confidence
92% confidence
Finding
This command repeats the same unpinned `npx skills` pattern, exposing users to registry or dependency compromise at install time. The surrounding context increases risk because the document is explicitly telling users how to add executable sub-skills, making copy-paste execution likely.

Skill Enumeration

Medium
Category
Agent Snooping
Content
- The user talks about "worlds/universes/spaces/scene settings".
- They want to browse gameplay and content organized by spaces/activities.

See `skills/neta-space/SKILL.md` for full details.

### 2. Content creation: `neta-creative`
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.

Skill Enumeration

Medium
Category
Agent Snooping
Content
- They want to create based on character settings or stories.
- They want to analyze the creative intent behind an existing work.

See `skills/neta-creative/SKILL.md` for full details.

### 3. Community browsing and interactions: `neta-community`
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.

Skill Enumeration

Medium
Category
Agent Snooping
Content
- The user says "show me what people are doing", "scroll the feed".
- They want to like or interact with specific works.

See `skills/neta-community/SKILL.md` for full details.

### 4. Research and recommendation engine: `neta-suggest`
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.

Skill Enumeration

Medium
Category
Agent Snooping
Content
- They want to understand popular tags/category structure/content distribution.
- They need systematic research before creating content.

See `skills/neta-suggest/SKILL.md` for full details.

### 5. Character creation and management: `neta-character`
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.

Skill Enumeration

Medium
Category
Agent Snooping
Content
- The user wants to "modify character settings", "update character backstory", or "change character appearance".
- The user wants to "list my characters" or "search for characters".

See `skills/neta-character/SKILL.md` for full details.

### 6. Elementum (visual style/concept) creation and management: `neta-elementum`
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.

Skill Enumeration

Medium
Category
Agent Snooping
Content
- The user wants to create reusable visual concepts (scenes, props, clothing, poses, atmospheres, memes).
- The user wants to "list my Elementa" or "search for Elementa".

See `skills/neta-elementum/SKILL.md` for full details.

## Migration notes (from legacy neta skill)
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.

Static analysis

No suspicious patterns detected.