Versioned security assessment

Report ID: SA-0919AC42

7/8/2026, 12:17:30 AM

nano-banana-2 security assessment v5

Skill Security Certification Report

Audit History
Audit model: codex Historical report
Skill name
nano-banana-2
Version
v5
Maintainer
agentspace-so
Source
github.com/agentspace-so/runcomfy-agent-skills/tree/main/nano-banana-2/
Coverage
1 Files scanned · 207 Lines analyzed
Policy version
Unavailable

Confirmed finding summary

No confirmed security findings

The completed audit recorded no confirmed security findings. This is not proof that the Skill has no side effects.

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 hits are markdown backticks, tables, or documentation links and are false positives. The skill intentionally installs or runs the RunComfy CLI and sends prompts to the RunComfy Model API. No prompt injection or malicious exfiltration intent was found in SKILL.md.

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

1 Files scanned · 207 Lines analyzed

11 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

    Binding unavailable

  2. Artifact

    Identity incomplete

  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.

Not recorded by this audit

Network access

May connect to external services.

Observed in 4 evidence locations

Filesystem access

May read or write local files.

Observed in 2 evidence locations

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 40 evidence locations

Capability review items (11)
Medium
SKILL.md:13
Ruby/shell backtick execution
`runcomfy run google/nano-banana-2/text-to-image` through the local
The skill explicitly says it calls the RunComfy text-to-image CLI through the local machine. This is intended external command execution that may transmit prompts and write output files.
Medium
SKILL.md:27-29
Ruby/shell backtick execution
```bash
The fenced bash block includes a global skill installation command through npx. This is user-directed, but it downloads and executes community package tooling.
Medium
SKILL.md:50-51
Ruby/shell backtick execution
1. **RunComfy CLI** — `npm i -g @runcomfy/cli`
The prerequisite instructs a global installation of the RunComfy CLI. Installing a third-party CLI changes the local environment and warrants review.
Medium
SKILL.md:51-52
Ruby/shell backtick execution
2. **RunComfy account** — `runcomfy login` opens a browser device-code flow.
The setup step runs the RunComfy login command and opens an authentication flow. It is expected setup, but it touches account credentials.
Medium
SKILL.md:76-80
Ruby/shell backtick execution
```bash
The bash example runs the RunComfy CLI with user prompt input and an output directory. It invokes a third-party CLI and can create local files.
Medium
SKILL.md:84-93
Ruby/shell backtick execution
```bash
The batch example runs the RunComfy CLI with prompt JSON and image settings. It performs external command execution for generated outputs.
Medium
SKILL.md:97-106
Ruby/shell backtick execution
```bash
The final-generation example runs the RunComfy CLI with a seed and output directory. It is intentional local command execution.
Medium
SKILL.md:110-117
Ruby/shell backtick execution
```bash
The web-grounded example runs the RunComfy CLI and enables web search. That combines external command execution with extra remote data retrieval.
Medium
SKILL.md:198
Ruby/shell backtick execution
The skill invokes `runcomfy run google/nano-banana-2/text-to-image` with a JSON body matching the sc
The text states that the CLI posts to the RunComfy Model API, polls, downloads outputs, and writes to the output directory. This confirms networked CLI execution.
Medium
SKILL.md:202
Ruby/shell backtick execution
- **Token storage**: `runcomfy login` writes the API token to `~/.config/runcomfy/token.json` with m
The line describes the RunComfy login command writing an API token into a local config file. This is expected, but it is credential handling by an external CLI.
Low
SKILL.md:198
Hardcoded URL
The skill invokes `runcomfy run google/nano-banana-2/text-to-image` with a JSON body matching the sc
The line states that the CLI posts to a RunComfy API URL and downloads results from RunComfy domains. This is the intended outbound network dependency and data-transfer risk.

Risk findings

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

No confirmed security findings were recorded for this completed audit.

Remediation

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

  1. FIX-001
    Medium
    External CLI execution
    Ask for user confirmation before installing packages or running RunComfy commands, especially with custom output directories.
  2. FIX-002
    Medium
    Credential handling
    Keep token storage documented, require owner-only permissions, and avoid printing or embedding tokens in prompts or examples.
  3. FIX-003
    Low
    Third-party prompt transmission
    Warn users that prompts and optional web search are sent to RunComfy and may include sensitive content.

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