Skills background-removal
π¦
background-removal
v1.2.0 Content revision r1 High Risk π Network accessπ Env variablesβοΈ External commandsπ Filesystem access
Remove Image Backgrounds with Verging AI
Background cleanup for products, profiles, and creative assets can take repeated manual editing. This skill sends images to Verging AI and returns transparent PNG results.
Supports: Claude Codex Code(CC)
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 "background-removal" from https://skillstore.io/skills/verging-ai-background-removal.md and its manifest at https://skillstore.io/api/skills/verging-ai-background-removal/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.
Test it
Using "background-removal". A local product photo named product.jpg with a cluttered table behind the item.
Expected outcome:
The skill reports a completed job, provides a transparent PNG result link, and saves the file when download is requested.
Using "background-removal". A remote portrait image URL that should become a profile cutout.
Expected outcome:
The skill downloads the source image, submits it to Verging AI, and returns the final transparent PNG link.
Security Audit
High Riskv5 β’ 7/7/2026 Open versioned report
Most static shell backtick alerts are Markdown fences or inline examples, but the skill genuinely performs curl-based network calls, uses VERGING_API_KEY, writes predictable files under /tmp, and uploads images to verging.ai. I found no prompt injection language. The remaining concerns are credential handling, arbitrary remote URL fetching, third-party image upload privacy, and predictable temporary paths.
2
Files scanned
166
Lines analyzed
19
Review items
0
False positives ignored
Confirmed security concerns (2)
High
Unvalidated Remote URL Fetch
The skill accepts image URLs and instructs curl -L to download the supplied URL before validation. Without scheme, host, and private-network checks, prompt-controlled input could cause unintended outbound requests from the agent environment.
The command format allows URLs, and the execution flow explicitly downloads an arbitrary URL with curl. The file gives no allowlist, scheme restriction, or private IP filtering guidance.
Medium
Third-Party Image Upload Privacy Risk
The skill uploads user images to the Verging AI API and notes that create-job handles R2 upload internally. Sensitive images may leave the local environment, and the skill does not describe retention, consent, or data handling limits.
The API flow directly posts the image file and states that server-side R2 upload is handled internally. This is disclosed functionality, but privacy impact is real and not covered by the static categories alone.
Capability review items (19)
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
Generic API/secret keys
- VERGING_API_KEY
The metadata declares VERGING_API_KEY as a required environment variable. That grants the skill access to a user secret and should be treated as real env access.
High
Generic API/secret keys
primaryEnv: VERGING_API_KEY
The metadata marks VERGING_API_KEY as the primary environment variable. This confirms the skill is designed to consume a secret credential.
High
Generic API/secret keys
| --api-key | -k | API Key | $VERGING_API_KEY |
The options table says the API key defaults to VERGING_API_KEY. This is legitimate design, but it confirms secret handling by the skill.
High
Generic API/secret keys
curl -H "Authorization: ApiKey $VERGING_API_KEY" \
The curl command places VERGING_API_KEY into an Authorization header for a remote API request. That is direct use of a user secret in a network command.
High
Generic API/secret keys
-H "Authorization: ApiKey $VERGING_API_KEY" \
The image upload command sends VERGING_API_KEY in an Authorization header. This is core functionality, but it is real credential transmission to a third-party service.
High
Generic API/secret keys
curl -H "Authorization: ApiKey $VERGING_API_KEY" \
The job polling command sends VERGING_API_KEY in an Authorization header. This confirms repeated use of the secret during remote API calls.
Medium
Ruby/shell backtick execution
```bash
The bash block instructs users or agents to run curl against verging.ai with an API key placeholder. This is an intended external command path and should remain visible as shell and network risk.
Medium
Ruby/shell backtick execution
```bash
The code block tells the agent to run curl with VERGING_API_KEY to check account credits. It executes an external command and sends a secret-bearing header to a remote API.
Medium
Ruby/shell backtick execution
```bash
The bash block posts an image file to the Verging AI API using curl and VERGING_API_KEY. It is legitimate functionality, but it is still external command execution with user data upload.
Medium
Ruby/shell backtick execution
```bash
The bash block instructs curl polling of a remote job endpoint with VERGING_API_KEY. This is expected behavior, but it is a real external command and network action.
Medium
Ruby/shell backtick execution
2. **Download remote image** (if URL): `curl -L -o /tmp/verging-bg-removal/photo.jpg "URL"`
The execution flow tells the agent to run curl -L against a user-supplied URL and write to /tmp. That is a real external command with network and file-system effects.
Medium
Temp directory access
-F "image=@/tmp/verging-bg-removal/photo.jpg" \
The multipart upload example reads an image from a fixed /tmp path. Predictable temp paths can expose stale or overwritten files if cleanup and permissions are weak.
Medium
Temp directory access
2. **Download remote image** (if URL): `curl -L -o /tmp/verging-bg-removal/photo.jpg "URL"`
The execution flow writes remote images to a fixed /tmp/verging-bg-removal/photo.jpg path. This creates file collision and leftover data risks in shared or repeated sessions.
Medium
Temp directory access
- **Temp directory:** `/tmp/verging-bg-removal/`
The skill explicitly defines /tmp/verging-bg-removal as its temp directory. This is expected, but still requires safe per-run naming and cleanup controls.
Low
Hardcoded URL
curl -H "Authorization: ApiKey vrg_sk_your_key_here" https://verging.ai/api/v1/auth/me
The curl example sends a request to the hardcoded Verging AI auth endpoint. The destination is disclosed and expected, but it is a real external network dependency.
Low
Hardcoded URL
curl -H "Authorization: Bearer vrg_sk_your_key_here" https://verging.ai/api/v1/auth/me
The alternative curl example also contacts the hardcoded Verging AI auth endpoint. This is transparent functionality, but confirms network use.
Low
Hardcoded URL
https://verging.ai/api/v1/auth/me
The check-credits command contacts the Verging AI auth endpoint. It is expected behavior, but it is still an external network request.
Low
Hardcoded URL
curl -X POST https://verging.ai/api/v1/background-removal/create-job \
The create-job command posts image data to a hardcoded Verging AI API endpoint. This confirms third-party network transfer as part of core functionality.
Low
Hardcoded URL
"https://verging.ai/api/v1/background-removal/jobs?job_ids=456"
The job polling command contacts a hardcoded Verging AI endpoint. The endpoint is expected, but network access is real.
Risk Factors
π Network access (8)
π Env variables (8)
βοΈ External commands (21)
SKILL.md:24-26 SKILL.md:26-39 SKILL.md:39-41 SKILL.md:41-47 SKILL.md:47-54 SKILL.md:54-57 SKILL.md:57-58 SKILL.md:58-61 SKILL.md:61-69 SKILL.md:69-70 SKILL.md:70-73 SKILL.md:73-76 SKILL.md:76-77 SKILL.md:77-79 SKILL.md:79 SKILL.md:84-87 SKILL.md:87-93 SKILL.md:93 SKILL.md:94-95 SKILL.md:95 SKILL.md:98-102
π Filesystem access (3)
Audited by: codex View Audit History β
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.
Copy report link
https://skillstore.io/skills/verging-ai-background-removal/audits/5?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_reportMarkdown badge
[](https://skillstore.io/skills/verging-ai-background-removal?utm_source=security_passport_badge)HTML badge
<a href="https://skillstore.io/skills/verging-ai-background-removal?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/verging-ai-background-removal/security.svg" alt="Skillstore security assessment" loading="lazy"></a>Embed card
<iframe src="https://skillstore.io/embed/skills/verging-ai-background-removal.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 Β· BibTeX Β· CFF)
APA citation
verging.ai. (2026). background-removal security audit report (audit version 5) [Author version 1.2.0]. Skillstore. https://skillstore.io/skills/verging-ai-background-removal/audits/5BibTeX citation
@techreport{verging-ai-verging-ai-background-removal-2026,
author = {verging.ai},
title = {background-removal security audit report (audit version 5)},
institution = {Skillstore},
year = {2026},
number = {5},
url = {https://skillstore.io/skills/verging-ai-background-removal/audits/5},
note = {Author version 1.2.0}
}CITATION.cff
cff-version: 1.2.0
message: "If you use this Skill, cite its author and this versioned security audit report."
title: "background-removal security audit report (audit version 5)"
version: "1.2.0"
type: report
authors:
- name: "verging.ai"
date-released: "2026-07-07"
url: "https://skillstore.io/skills/verging-ai-background-removal/audits/5"
identifiers:
- type: other
value: "skillstore:verging-ai-background-removal:audit:5"
description: "Skillstore immutable audit report identifier"
Compare variants
2 installable variantsEach author remains a separate installable skill. The recommended variant is ranked by Skillstore evidence.
Why this variant is first
Highest Skillstore Score
Author / source Skillstore Score Evidence Confidence Skillstore usage Updated
inference-sh Recommended
inference-sh-background-removal
Skillstore Score 50
Evidence Confidence Medium
Skillstore usage 7
Updated
2026-08-21
verging.ai Current
verging-ai-background-removal
Skillstore Score 38
Evidence Confidence Medium
Skillstore usage 6
Updated
2026-08-21
Skillstore Score
Why this score Evidence Confidence: Medium 55
Architecture
100
Maintainability
87
Content
67
Community
91
Spec Compliance
What You Can Build
Prepare Product Photos
Create transparent product images for catalog pages, ads, and marketplace listings.
Create Design Cutouts
Remove busy backgrounds before placing people, objects, or assets into design compositions.
Automate Asset Cleanup
Run background removal inside agent workflows that prepare images for documentation or campaigns.
Try These Prompts
Remove a Local Background
Remove the background from ./photo.jpg and show me the transparent PNG result link.
Save the Transparent Output
Remove the background from ./headshot.png and save the transparent PNG as ./headshot-transparent.png.
Process a Remote Image
Remove the background from https://example.com/photo.png and download the transparent PNG to the current folder.
Validate and Run
Check my Verging AI credits, confirm ./product.webp is under 10 MB, then remove its background and save the result.
Best Practices
- Use trusted image sources and review remote URLs before downloading them.
- Keep VERGING_API_KEY in the environment and avoid sharing command logs that include headers.
- Use explicit output paths and delete temporary files after processing sensitive images.
Avoid
- Do not process confidential images without approval to send them to an external service.
- Do not pass untrusted remote URLs without checking the destination first.
- Do not reuse downloaded temporary files as if they were verified originals.
Frequently Asked Questions
Does this skill need an API key?
Yes. It uses VERGING_API_KEY to authenticate with the Verging AI API.
Does the background removal run locally?
No. The skill uploads images to Verging AI and receives a transparent PNG result.
Which image formats are supported?
It supports JPG, PNG, and WebP images up to 10 MB.
How many credits does one image use?
The skill documentation states that each image costs one credit.
Can it process a remote image URL?
Yes. It can download a remote image first, but users should verify the URL before processing.
Can it save the transparent PNG file?
Yes. Use an output path or download option to save the generated transparent PNG.
Developer Details
Author
verging.aiLicense
MIT
Author version
v1.2.0
Skillstore revision
r1
Ref
36e07d5e13068e5be64447e8f20b427cf2cbd21a
Maintenance freshness
7/18/2026
Usage
1 downloads Β· 0 views