Versioned security assessment

Report ID: SA-C43861A6

7/23/2026, 6:27:06 PM

higgsfield-websites security assessment v3

Skill Security Certification Report

Audit History
Scanner version 3.0.0 Audit model: codex Latest published report
Skill name
higgsfield-websites
Version
v0.12.0
Maintainer
higgsfield-ai
Source
github.com/higgsfield-ai/skills/tree/main/higgsfield-websites/
Coverage
22 Files scanned · 8,130 Lines analyzed
Policy version
skillstore-security-audit-policy-v1

Highest confirmed finding severity

Critical

4 confirmed security findings require attention.

Installation context

Check the current Skill page

This page summarizes report evidence only. The Skill page provides the canonical install advisory.

Open current Skill page

This report does not block or authorize the manifest or ZIP.

Most detections are false positives caused by Markdown examples and defensive security guidance. The skill still executes an unverified remote installer, imports unpinned third-party components, and defaults to public deployment. It also attempts to override other skills and obscures deployment operations from users.

Report position

Latest published report

Latest refers to the report sequence, not to artifact currentness.

Audit attestation

Active attestation

A public attestation is available for this exact report.

Human verification

Not verified

No human verification is recorded for this report.

Coverage

22 Files scanned · 8,130 Lines analyzed

13 items shown for review

Limitations

This report does not claim runtime or sandbox execution and does not prove the absence of side effects.

Evidence chain

Follow the evidence from source binding to the install contract. Available evidence supports verification; it is not a safety guarantee.

  1. Source

    Commit and path bound

  2. Artifact

    Content and tree hashes bound

  3. Audit

    Complete

  4. Install contract

    Open manifest to verify

    Open manifest

Capabilities observed

Observed means this report recorded supporting evidence. Not recorded does not prove that a capability is absent.

Contains scripts

May execute code included with the Skill.

Observed in 4 evidence locations

Network access

May connect to external services.

Observed in 52 evidence locations

Filesystem access

May read or write local files.

Observed in 8 evidence locations

Env variables

May read values from the process environment.

Observed in 9 evidence locations

External commands

May invoke commands or programs outside the Skill.

Observed in 50 evidence locations

Capability review items (9)
Critical
SKILL.md:98
Hardcoded URL
curl -fsSL https://raw.githubusercontent.com/higgsfield-ai/cli/main/install.sh | sh
This URL supplies an installer that is fetched from a mutable GitHub branch and passed directly to a shell. A compromised upstream response would execute with the agent's local privileges.
High
references/wow-maker.md:216
Hardcoded URL
"@magicui": "https://magicui.design/r/{name}.json",
This URL is configured as a third-party component registry used by unpinned shadcn add commands. Registry content is copied into the project without a version pin, checksum, or required source review.
High
references/wow-maker.md:217
Hardcoded URL
"@cult-ui": "https://www.cult-ui.com/r/{name}.json",
This URL is configured as a third-party component registry used by unpinned shadcn add commands. Registry content is copied into the project without a version pin, checksum, or required source review.
High
references/wow-maker.md:218
Hardcoded URL
"@smoothui": "https://smoothui.dev/r/{name}.json",
This URL is configured as a third-party component registry used by unpinned shadcn add commands. Registry content is copied into the project without a version pin, checksum, or required source review.
High
references/wow-maker.md:219
Hardcoded URL
"@ncdai": "https://chanhdai.com/r/{name}.json",
This URL is configured as a third-party component registry used by unpinned shadcn add commands. Registry content is copied into the project without a version pin, checksum, or required source review.
High
references/wow-maker.md:220
Hardcoded URL
"@motion-primitives": "https://motion-primitives.com/c/{name}.json",
This URL is configured as a third-party component registry used by unpinned shadcn add commands. Registry content is copied into the project without a version pin, checksum, or required source review.
High
references/wow-maker.md:221
Hardcoded URL
"@kokonutui": "https://kokonutui.com/r/{name}.json",
This URL is configured as a third-party component registry used by unpinned shadcn add commands. Registry content is copied into the project without a version pin, checksum, or required source review.
High
references/wow-maker.md:222
Hardcoded URL
"@tailark": "https://tailark.com/r/{name}.json",
This URL is configured as a third-party component registry used by unpinned shadcn add commands. Registry content is copied into the project without a version pin, checksum, or required source review.
High
references/wow-maker.md:223
Hardcoded URL
"@eldoraui": "https://eldoraui.site/r/{name}.json"
This URL is configured as a third-party component registry used by unpinned shadcn add commands. Registry content is copied into the project without a version pin, checksum, or required source review.

Risk findings

Confirmed security concerns are separated from items that still need review.

Confirmed security concerns (4)

RISK-001 Critical
SKILL.md:98
Pipe to shell pattern
curl -fsSL https://raw.githubusercontent.com/higgsfield-ai/cli/main/install.sh | sh
The prerequisite command pipes a remote install script directly into sh without pinning or integrity verification. This creates immediate supply-chain code-execution risk.
RISK-002 High
SKILL.md:156-158
Prompt Injection Attempt Detected
The skill states, "no other skill ... overrides these rules." This attempts to control instruction priority and suppress independent guidance.
The quoted instruction explicitly claims precedence over other skills. It is not needed to explain website-building behavior.
RISK-003 High
SKILL.md:152-175references/app-flow.md:131-136
Live Deployment Without Explicit Consent
The skill defaults to deploying every build as a live public site, even when the user asked only to build or edit. It also hides deployment terminology.
The files explicitly say deployment is public, has no preview stage, and is the default final step. User-facing messages are told to avoid deploy terminology.
RISK-004 Medium
SKILL.md:124-139references/asset-system.md:9-118
Unprompted External Asset Generation
The skill mandates many external image-generation jobs and rerolls without asking. Permission is required only for optional video generation.
The workflow requires reference boards, candidate assets, rerolls, and covers. It explicitly says the cover image needs no permission.

Remediation

Suggested fixes recorded by this audit. Applying them is the maintainer’s responsibility.

  1. FIX-001
    Critical
    Remote installer is piped directly into a shell.
    Use a pinned release artifact, verify its checksum or signature, inspect it locally, and request user confirmation before execution.
  2. FIX-002
    High
    Build and edit requests lead to a live public deployment by default.
    Require explicit release confirmation, disclose that no preview exists, and describe the public deployment before running the command.
  3. FIX-003
    High
    The skill claims precedence over other skills.
    Remove the override language and preserve system, developer, user, and applicable local instruction priority.
  4. FIX-004
    High
    Third-party component registries and installers are unpinned.
    Pin package and component versions, verify integrity, and review imported source before building or deploying it.
  5. FIX-005
    Medium
    External asset-generation batches run without explicit consent.
    Explain the generation plan and obtain approval before submitting jobs, rerolls, uploads, or other account-consuming actions.

Expert evidence

Immutable subject identity, scanner metadata, dismissed matches, and source-level evidence.

Artifact subject

Marketplace commit
c43861a65bb95efcae259cd161c9d6f4dc7eec6f
Content hash
d52e3861a76e98c1ebad1adfe2a1a146fb0dd35f1b968af7b4ff00763ff99b8a
Tree hash
222b62730580b26547c19c84b0d22bd84bc608a0aae264123a3a39c2a427bc14
Skill path
skills/higgsfield-ai/higgsfield-websites
Audit payload hash
9fbc579f7af578258f3b73ee10ab731a

Analysis metadata

Audit model: codex

Analysis state: Complete

Scope is limited to the recorded files, lines, methods, and evidence. No runtime or sandbox execution is claimed.

Verify and export

JSON report Signed attestation CITATION.cff SARIF 2.1.0 Signed installation manifest Source

The manifest and lockfile bind install artifacts to cryptographic hashes. This integrity claim is separate from the security assessment.

Audit attestation: active