Back to skill

Security audit

hyperframes-registry

Security checks for vulnerabilities and agentic risk

Overview

ClawScan rates this skill suspicious with medium confidence because scanners report undisclosed external actions and unapproved sharing of search details beyond its catalog and installation purpose.

Consider the suspicious verdict and the reported external sharing and unpinned package commands before installing. The skillspector scan was partial: it covered all 13 listed components but could not resolve one local path-like reference, so the evidence is incomplete. The separate AIG scan completed with no findings, which does not negate the skillspector concerns.

Vulnerability Patterns
  • 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
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
Findings (44)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding

Rendering previews and taking QA snapshots are reasonable supporting activities for visual components. However, the npx hyperframes publish command introduces an external publishing capability beyond the declared component catalog and installation workflow. The snippet documents this command; it does not demonstrate that publishing runs automatically.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
90% confidence
Finding

Formatting components and regenerating registry/catalog data are supporting implementation details consistent with maintaining the component collection. However, the supplied shell workflow also performs outbound publishing and remote repository contribution operations, which are not disclosed by the description's component discovery and installation purpose. These commands appear in contribution documentation; the chunk does not establish that they execute automatically.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
90% confidence
Finding

The substantive issue is not merely an abbreviated description: the skill directs the agent to report unsuccessful searches with the query and desired effect before proceeding, without requiring user approval. Users are told search is local, so this subsequent feedback step can unexpectedly disclose confidential project requirements outside the local search workflow.

Content

No source excerpt is available for this finding.

Anti-Refusal Statement

High
Category
Anti-Refusal
Confidence
80% confidence
Finding

Skill instructs the agent to omit warnings, disclaimers, or ethical commentary. Stripping safety caveats hides risk from the user and is a common jailbreak preamble.

Content

Scanner excerpt · references/component-quality-bar.md (reported line 33)May include surrounding context.

md
| **Visual**     | render and eye | renders at all, implements its description, legible, deterministic                       |

`hyperframes check` is not a visual gate. It passes compositions that render nothing: a blank plot
produces no error, no warning and no layout finding, because an empty render is a valid render. No
criterion below may rest on `check` alone.

### The mount harness

MCP server referenced without pinned version: 'npx hyperframes'.

Medium
Category
MCP Rug Pull
Confidence
70% confidence
Finding

npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx hyperframes'.

Medium
Category
MCP Rug Pull
Confidence
70% confidence
Finding

npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx hyperframes'.

Medium
Category
MCP Rug Pull
Confidence
70% confidence
Finding

npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx hyperframes'.

Medium
Category
MCP Rug Pull
Confidence
70% confidence
Finding

npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx hyperframes'.

Medium
Category
MCP Rug Pull
Confidence
70% confidence
Finding

npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx hyperframes'.

Medium
Category
MCP Rug Pull
Confidence
70% confidence
Finding

npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.

Content

No source excerpt is available for this finding.

The skill requires an automatic, potentially account-linked telemetry beacon unrelated to finding or installing visual components.

Medium
Category
Excessive Agency
Confidence
97% confidence
Finding

The skill directs an automatic usage beacon on activation, potentially associating skill use with a signed-in account without task-specific approval. Disclosure and an existing opt-out reduce the concern, and no project-content transfer is shown, but this remains an unnecessary external side effect for a catalog workflow.

Content

No source excerpt is available for this finding.

The search-miss workflow submits query and effect details as feedback without a privacy disclosure.

Medium
Category
Output Handling
Confidence
93% confidence
Finding

The prescribed feedback command includes the user's search query and requested effect, which can contain confidential client or project details. Unlike the optional on-device download, this workflow has no explicit approval gate or explanation of destination and retention; it is the same underlying feedback-disclosure issue identified by TP4.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx hyperframes'.

Medium
Category
MCP Rug Pull
Confidence
87% confidence
Finding

This is a CLI invocation, not an MCP server reference, but the unpinned execution risk is valid: if a suitable local installation is unavailable, npx can fetch and execute an unreviewed release. A compromised package could execute with the contributor's permissions; publishing previews is legitimate, but does not require accepting this supply-chain exposure.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx oxfmt'.

Medium
Category
MCP Rug Pull
Confidence
87% confidence
Finding

Although oxfmt is a formatter rather than an MCP server, npx can download and execute it when it is not installed locally. Without a demonstrated locked dependency, this can expose repository files and contributor credentials to a compromised release; the routine formatting purpose does not restrict the executable's permissions.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx tsx'.

Medium
Category
MCP Rug Pull
Confidence
85% confidence
Finding

The generation script is local, but its unpinned tsx runner can be fetched by npx if no suitable local installation exists. A compromised runner could execute arbitrary code before or while running the legitimate script; this is a conditional package supply-chain issue, not an MCP configuration issue.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx tsx'.

Medium
Category
MCP Rug Pull
Confidence
85% confidence
Finding

This repeats the unpinned tsx execution risk for catalog generation: npx can fetch a runner if it is absent locally. The local script path does not establish trust in the runner, so a compromised release could access or modify the checkout and other resources available to the contributor.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx hyperframes'.

Medium
Category
MCP Rug Pull
Confidence
87% confidence
Finding

The shipping workflow repeats an unpinned npx hyperframes invocation, which can execute a newly fetched package when a suitable local installation is unavailable. Publishing is an expected workflow action rather than evidence of malicious exfiltration, but a compromised CLI would run in a potentially authenticated contributor environment.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx oxfmt'.

Medium
Category
MCP Rug Pull
Confidence
85% confidence
Finding

The quality checklist repeats the same unpinned formatter invocation rather than introducing a separate attack mechanism. The --check option limits intended formatting behavior, but does not sandbox a compromised package fetched by npx.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx hyperframes'.

Medium
Category
MCP Rug Pull
Confidence
85% confidence
Finding

This checklist item repeats the previously identified unpinned publishing command. If npx fetches the package rather than resolving a trusted local installation, a compromised release could execute with contributor permissions; no malicious intent or actual compromise is established by the text.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx hyperframes'.

Medium
Category
MCP Rug Pull
Confidence
96% confidence
Finding

This is a CLI invocation, not an MCP server reference, but the unpinned-package concern applies. If no suitable local package exists, npx can fetch and execute a mutable registry release with the agent's permissions; a compromised release could access workspace files and credentials even though the requested operation only lists catalog entries.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx hyperframes'.

Medium
Category
MCP Rug Pull
Confidence
96% confidence
Finding

The block filter does not mitigate the risk of executing an unpinned HyperFrames CLI through npx. Although this is ordinary discovery documentation rather than an MCP configuration, a compromised fetched release could execute code with the agent's privileges.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx hyperframes'.

Medium
Category
MCP Rug Pull
Confidence
96% confidence
Finding

Selecting component entries still executes the same unpinned CLI, which npx may obtain from a mutable registry release. The intended read-only operation does not constrain what compromised package code could do with the agent's filesystem or credentials.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx hyperframes'.

Medium
Category
MCP Rug Pull
Confidence
96% confidence
Finding

The type and tag arguments only narrow catalog results; they do not establish trust in the executable package. Using unpinned npx can therefore expose the agent to a compromised future release despite the benign discovery purpose.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx hyperframes'.

Medium
Category
MCP Rug Pull
Confidence
96% confidence
Finding

JSON output may be deterministic for a given CLI and registry state, but it does not pin or authenticate the CLI being executed. Recommending this unpinned command for agents and CI increases exposure because those environments may contain repository credentials or other secrets.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx hyperframes'.

Medium
Category
MCP Rug Pull
Confidence
96% confidence
Finding

This unpinned npx invocation can execute a compromised package release with the agent's permissions. Unlike the other catalog examples, the documented picker also installs the selected item immediately, adding a workspace-modification step to discovery.

Content

No source excerpt is available for this finding.