Versioned security assessment

Report ID: SA-C67D8537

7/6/2026, 3:28:24 PM

higgsfield-soul-id security assessment v3

Skill Security Certification Report

Audit History

Audit model: codex Historical report

Skill name: higgsfield-soul-id
Version: v3
Maintainer: higgsfield-ai
Source: github.com/higgsfield-ai/skills/tree/main/higgsfield-soul-id/
Coverage: 3 Files scanned · 147 Lines analyzed
Policy version: Unavailable

Highest confirmed finding severity

Critical

2 confirmed security findings require attention.

Installation context

Historical evidence

This report may not describe the currently installable artifact. Open the current Skill page for install guidance.

Open current Skill page

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

Most static findings are false positives from Markdown backticks, CLI flags, or product reference IDs. The audit confirms the remote installer pipe-to-shell command, the related network dependency, command-construction risks, and sensitive biometric photo handling. No prompt injection attempt was found in the reviewed files.

Report position

Historical report

Open audit history before using this report to install.

Audit attestation

Not attestable

The required immutable binding is incomplete.

Human verification

Not verified

No human verification is recorded for this report.

Coverage

3 Files scanned · 147 Lines analyzed

7 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.

Source

Binding unavailable
Artifact

Identity incomplete
Audit

Complete
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.

Not recorded by this audit

Network access

May connect to external services.

Observed in 1 evidence location

Filesystem access

May read or write local files.

Not recorded by this audit

Env variables

May read values from the process environment.

Not recorded by this audit

External commands

May invoke commands or programs outside the Skill.

Observed in 24 evidence locations

Capability review items (5)

Critical

SKILL.md:28-30

Ruby/shell backtick execution

```bash

The fenced block contains a command that downloads an installer and pipes it to sh. This is shell execution of remote code without pinning or verification.

Medium

SKILL.md:50-53

Ruby/shell backtick execution

```bash

The command block builds a Higgsfield CLI call from user-provided name and image path values. Without explicit argument escaping, those values could be interpreted by the shell before the CLI receives them.

Medium

SKILL.md:53-55

Ruby/shell backtick execution

```

This range states that the CLI auto-uploads local paths and then waits on the created job. It is expected for this skill, but it is still external command execution that transfers sensitive local files.

Medium

SKILL.md:62-65

Ruby/shell backtick execution

```bash

The command examples include prompt and Soul ID placeholders that an agent could interpolate into Bash. They should be passed as escaped arguments to avoid shell interpretation of user text.

Medium

SKILL.md:29

Hardcoded URL

curl -fsSL https://raw.githubusercontent.com/higgsfield-ai/cli/main/install.sh | sh

The line fetches an installer from a hardcoded GitHub URL at runtime. The URL targets a mutable main-branch script and is paired with shell execution.

Risk findings

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

Confirmed security concerns (2)

RISK-001 Critical

SKILL.md:29

Pipe to shell pattern

curl -fsSL https://raw.githubusercontent.com/higgsfield-ai/cli/main/install.sh | sh

Line 29 pipes a remote install script from GitHub directly into sh. That can execute changed or compromised network content with the user privileges of the agent session.

RISK-002 Medium

SKILL.md:5-6 SKILL.md:43-54 references/photo-guide.md:7-29

Sensitive Biometric Data Handling

The skill asks for 5 to 20 face photos and auto-uploads local paths to train a reusable identity model, but it does not require consent or privacy confirmation.

The cited lines directly describe face-faithful identity training, photo collection, and automatic upload of local paths. No nearby instruction requires consent, ownership verification, or privacy disclosure.

Remediation

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

FIX-001
Critical
Remote installer is piped directly into a shell.

Replace the curl-to-sh installer with a pinned release, checksum verification, or package-manager install steps that require explicit user confirmation.
FIX-002
Medium
CLI examples interpolate user-provided names, prompts, paths, and IDs into shell commands.

Require argument-array execution or clear quoting and escaping rules before passing user values to Bash.
FIX-003
Medium
The workflow uploads face photos to train a reusable identity model without a consent or privacy preflight.

Add instructions to confirm photo ownership, subject consent, upload destination, and acceptable use before training starts.

Expert evidence

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

Artifact subject

Marketplace commit: Unavailable
Content hash: Unavailable
Tree hash: Unavailable
Skill path: Unavailable
Audit payload hash: Unavailable

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 CITATION.cff SARIF 2.1.0 Signed installation manifest

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

Audit attestation: not_attestable