Skills image-generation

📦

image-generation

Name: image-generation
Author: danielgwilson

Content revision r1 High Risk ⚙️ External commands🌐 Network access📁 Filesystem access

Generate Hosted Images With Agent Prompts

Agents need reliable image generation without local setup or provider keys. This skill guides Claude, Codex, and Claude Code through a hosted Image Skill CLI and API workflow.

Supports: Claude Codex Code(CC)

⚠️ 38 Poor

Stop for confirmation before installing.

Review the plan and obtain explicit user consent before changing files.

Install with my Agent

Copy this request to your Agent. It includes the canonical Skill page and manifest.

Agent request

Review the Skillstore skill "image-generation" from https://skillstore.io/skills/danielgwilson-image-generation.md and its manifest at https://skillstore.io/api/skills/danielgwilson-image-generation/manifest. Verify the artifact. Stop and obtain explicit user consent before installing or changing files.

Your Agent should still show its plan and request any confirmation required by the security policy.

Agent-readable resources

Use these links when an AI agent, crawler, or script needs clean context instead of reading the full page.

Markdown detail GET /skills/danielgwilson-image-generation.md Signed manifest GET /api/skills/danielgwilson-image-generation/manifest Signed lockfile GET /api/skills/danielgwilson-image-generation/lockfile

Test it

Using "image-generation". Generate a concept image of a compact field camera on a stainless workbench.

Expected outcome:

A hosted media link, model name, prompt summary, cost receipt, and recoverable job reference.

Using "image-generation". Check whether image generation is ready without spending money.

Expected outcome:

A readiness status, missing inputs if any, and the recommended no-spend next step.

Using "image-generation". Recover the last product image job and make it brighter.

Expected outcome:

The recovered media link, previous prompt summary, revised prompt direction, and new hosted output details.

Security Audit

High Risk

v4 • 7/5/2026 Open versioned report

The skill is not a prompt-injection attempt, but it relies on external hosted services and several shell workflows. Real risks include npx execution of remote packages, global skill installation, remote-supplied next commands, hidden agent-state paths, and paid media generation.

Files scanned

Lines analyzed

Review items

False positives ignored

Confirmed security concerns (2)

High

SKILL.md:28

Remote Command Delegation

The skill tells agents to run data.next_command returned by the hosted guide when data.next_command_copy_runnable is true. This lets a remote service influence shell commands an agent executes.

The instruction is explicit on line 28 and is separate from the static backtick pattern. The guard flag reduces accidental use but does not remove remote command influence.

Medium

SKILL.md:28 SKILL.md:39

Automated Paid Media Spend Workflow

The skill allows image creation when media spend is allowed and provides a maximum estimated cost per image. Ambiguous approval could let an agent trigger paid generation unexpectedly.

The skill includes spend gating language and a cost limit flag, so the monetary workflow is real. The risk depends on how the host enforces user approval.

Capability review items (12)

These are real local capabilities that may be expected for this skill, so they require review but are not counted as confirmed malicious behavior.

High

SKILL.md:46-48

Ruby/shell backtick execution

```bash

The command block at lines 46-48 installs a community skill globally for Codex with -g and -y. This can modify agent behavior without an interactive confirmation step.

High

SKILL.md:52-54

Ruby/shell backtick execution

```bash

The command block at lines 52-54 installs the canonical image-skill globally with automatic yes flags. A global skill install can alter future agent behavior and should require trust review.

Medium

SKILL.md:28

Ruby/shell backtick execution

Do not bring provider API keys, create provider accounts, run a local model server, or wire a separa

Line 28 instructs agents to run data.next_command returned by the hosted guide when a flag is true. That delegates command execution to remote service output, which is a real execution risk.

Medium

SKILL.md:32-34

Ruby/shell backtick execution

```bash

The fenced bash block at lines 32-34 contains an npx command that downloads and runs image-skill@latest. Running an unpinned remote package can execute arbitrary package code.

Medium

SKILL.md:38-40

Ruby/shell backtick execution

```bash

The bash block at lines 38-40 runs npx image-skill@latest with a prompt and cost parameter. It executes remote package code and sends generation requests to the hosted runtime.

Medium

SKILL.md:62-67

Ruby/shell backtick execution

```bash

The bash block at lines 62-67 changes HOME and CODEX_HOME, creates directories, and runs a global skill install. These are real shell operations affecting agent state.

Medium

SKILL.md:67-78

Ruby/shell backtick execution

```

The range includes an instruction on line 78 to run image-skill feedback create with attempted command details. It is a legitimate support workflow, but it still invokes an external CLI and can transmit operational data.

Medium

SKILL.md:59

Hidden file access

`$HOME/.agents`; `CODEX_HOME` keeps Codex profile state on the same writable

Line 59 states that the adapter writes to $HOME/.agents and uses CODEX_HOME for profile state. Hidden agent-state paths can affect future agent behavior and should be explicit.

Medium

SKILL.md:63

Hidden file access

export HOME="$PWD/.agent-home"

Line 63 redirects HOME to a hidden workspace directory. This changes where tools read and write user state, which is a meaningful filesystem side effect.

Medium

SKILL.md:64

Hidden file access

export CODEX_HOME="$HOME/.codex"

Line 64 redirects CODEX_HOME into $HOME/.codex. That changes Codex profile state storage and can influence later agent sessions.

Low

SKILL.md:11

Hardcoded URL

canonical_api_url: https://api.image-skill.com

The skill declares https://api.image-skill.com as the canonical API endpoint. Image prompts, job metadata, and media requests may be sent to this third-party hosted service.

Low

SKILL.md:76

Hardcoded URL

- Hosted API: https://api.image-skill.com

Line 76 identifies the hosted API endpoint used by the workflow. Calls to that service can expose prompts, job details, and media metadata to the provider.

Risk Factors

⚙️ External commands (13)

SKILL.md:24 SKILL.md:28 SKILL.md:32-34 SKILL.md:34-38 SKILL.md:38-40 SKILL.md:40-46 SKILL.md:46-48 SKILL.md:48-52 SKILL.md:52-54 SKILL.md:54-59 SKILL.md:59 SKILL.md:62-67 SKILL.md:67-78

🌐 Network access (8)

SKILL.md:5 SKILL.md:6 SKILL.md:10 SKILL.md:11 SKILL.md:73 SKILL.md:74 SKILL.md:75 SKILL.md:76

📁 Filesystem access (3)

SKILL.md:59 SKILL.md:63 SKILL.md:64

Audited by: codex View Audit History →

Share & cite this report

Share the versioned assessment report, neutral badge, embed card, and citations. Skillstore reports evidence without deciding whether this Skill is safe.

Open versioned report

Copy report link

https://skillstore.io/skills/danielgwilson-image-generation/audits/4?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_report

Markdown badge

[![Skillstore security assessment](https://skillstore.io/badges/skills/danielgwilson-image-generation/security.svg)](https://skillstore.io/skills/danielgwilson-image-generation?utm_source=security_passport_badge)

HTML badge

<a href="https://skillstore.io/skills/danielgwilson-image-generation?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/danielgwilson-image-generation/security.svg" alt="Skillstore security assessment" loading="lazy"></a>

Embed card

<iframe src="https://skillstore.io/embed/skills/danielgwilson-image-generation.html" title="Skillstore Security Assessment" sandbox="allow-popups allow-popups-to-escape-sandbox" loading="lazy" referrerpolicy="no-referrer" width="420" height="180"></iframe>

Academic citations (APA · BibTeX · CFF)

APA citation

danielgwilson. (2026). image-generation security audit report (audit version 4) [Author version unspecified]. Skillstore. https://skillstore.io/skills/danielgwilson-image-generation/audits/4

BibTeX citation

@techreport{danielgwilson-danielgwilson-image-generation-2026,
  author = {danielgwilson},
  title = {image-generation security audit report (audit version 4)},
  institution = {Skillstore},
  year = {2026},
  number = {4},
  url = {https://skillstore.io/skills/danielgwilson-image-generation/audits/4},
  note = {Author version unspecified}
}

CITATION.cff

cff-version: 1.2.0
message: "If you use this Skill, cite its author and this versioned security audit report."
title: "image-generation security audit report (audit version 4)"
version: "unspecified"
type: report
authors:
  - name: "danielgwilson"
date-released: "2026-07-05"
url: "https://skillstore.io/skills/danielgwilson-image-generation/audits/4"
identifiers:
  - type: other
    value: "skillstore:danielgwilson-image-generation:audit:4"
    description: "Skillstore immutable audit report identifier"

Skillstore Score

Why this score Evidence Confidence: Medium

Architecture

Maintainability

Content

Community

Spec Compliance

What You Can Build

Create Product Concept Images

Generate hosted visual concepts from product prompts, with model settings and cost receipts for review.

Add Image Output to Agent Workflows

Use repeatable CLI commands so agents can create, recover, and cite generated media during project work.

Draft Campaign Visuals Quickly

Create illustration or product-image drafts without setting up separate provider accounts or local model servers.

Try These Prompts

Create a Simple Image

Use the image-generation skill to create a hosted image of [subject] in [style]. Keep spend under [amount] and return the image URL and receipt.

Explore Style Options

Generate three prompt variations for [product or scene]. Use guide mode first, then create the best option only after I approve spend.

Produce Product Visuals

Create hosted product images for [product]. Use [model], [aspect ratio], and [budget], then summarize URLs, costs, and prompts.

Recover and Improve a Job

Find the previous Image Skill job for [project], recover the hosted media, create an improved version with [changes], and record feedback if a capability is missing.

Best Practices

Run guide mode first, and create images only after explicit spend approval.
Review remote commands before execution, especially npx and data.next_command.
Keep prompts, job identifiers, URLs, and receipts in the project record.

Avoid

Do not paste secrets or provider API keys into prompts or command arguments.
Do not run remote-supplied commands without reading and approving them.
Do not assume generated media is private or available offline.

Frequently Asked Questions

Does this require provider API keys?

No. The skill is designed around the hosted Image Skill runtime instead of direct provider credentials.

Can it work offline?

No. It needs network access to the hosted Image Skill service and npm package sources.

Which AI tools can use it?

The report lists Claude, Codex, and Claude Code as supported tools.

Can image generation cost money?

Yes. The workflow includes spend approval language and estimated cost parameters.

Where are generated images stored?

The skill describes durable hosted media URLs managed by the Image Skill service.

What should I review before installation?

Review the npx commands, global install flags, external API endpoints, and HOME or CODEX_HOME path changes.

Developer Details

Author

danielgwilson

License

MIT

Skillstore revision

Version notice

The author did not declare a version.

Repository

https://github.com/danielgwilson/image-skill-cli/tree/main/skills/image-generation

Ref

7e9e368dc1a370f2040369ebcd0db06d90cb48e5

Maintenance freshness

7/18/2026

Usage

1 downloads · 0 views

File structure

📄 SKILL.md