Skills skywork-search
📦
skywork-search
Content revision r1 High Risk ⚙️ External commands🌐 Network access📁 Filesystem access🔑 Env variables
Search the Web with Skywork
Current research requests need sources beyond model training data. This skill uses the Skywork API to search the web and save source-backed 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 "skywork-search" from https://skillstore.io/skills/skyworkai-skywork-search.md and its manifest at https://skillstore.io/api/skills/skyworkai-skywork-search/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 "skywork-search". Search for recent battery recycling policy updates.
Expected outcome:
A brief summary with source URLs, policy dates, and notes on claims that need verification.
Using "skywork-search". Find current documentation for a new API release.
Expected outcome:
- Official documentation URL and version notes.
- Relevant changelog entries.
- Known limitation notes from support pages.
Using "skywork-search". Research funding trends for climate technology.
Expected outcome:
A sourced overview with recent funding numbers, major investors, and publication links.
Security Audit
High Riskv5 • 7/7/2026 Open versioned report
The skill implements legitimate Skywork web search, but it has significant security exposure from token caching, outbound network requests, and external browser launch commands. I found no prompt injection language, but the search script builds output filenames from raw query text and can write outside the intended temporary directory.
3
Files scanned
477
Lines analyzed
21
Review items
0
False positives ignored
Confirmed security concerns (3)
High
Path Traversal in Search Result Filename
The search result path is built from the raw query string before writing the output file. A query containing slashes or an absolute path can write outside the temporary directory.
The command-line query is directly embedded in the output filename and then opened for writing. I found no sanitization, basename check, or path containment check.
High
Credential Cache Lacks Permission Hardening
The authentication module writes token data to a hidden home-directory file without visible permission restrictions. Default file permissions may allow unintended local access to the token.
The token save function opens the cache file and writes JSON token data. I found no chmod, opener mode control, or operating-system credential store use near the write.
Medium
Environment-Controlled Endpoints Receive Tokens
Environment variables can replace the authentication or search gateway endpoints. If misconfigured, token validation or search requests can send credentials to a non-default host.
The API base and gateway URL are read from the environment, and later requests include token data. This can be legitimate configuration, so confidence is high enough to flag but not absolute.
Capability review items (21)
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
Python subprocess.Popen
subprocess.Popen(["open", url], start_new_session=True,
The macOS branch starts an external browser opener with a login URL built from the configured base URL. It uses an argument list, but it still executes a local process and can direct the user to an external site.
High
Python subprocess.Popen
subprocess.Popen(["start", url], shell=True, start_new_session=True,
The Windows branch invokes subprocess.Popen with shell=True while passing a URL influenced by the login base configuration. That creates a higher command-injection and unsafe browser-launch risk.
High
Python subprocess.Popen
subprocess.Popen(["xdg-open", url], start_new_session=True,
The Linux branch starts xdg-open with a login URL built from the configured base URL. It is intended browser behavior, but it is still external process execution from a skill script.
High
Hidden file in home directory
Token is stored in user home directory as ~/.skywork_token (JSON format).
The docstring states that the token is stored as ~/.skywork_token. A hidden home-directory credential cache is a real sensitive-file persistence risk.
High
Hidden file in home directory
TOKEN_FILE = os.path.expanduser("~/.skywork_token")
TOKEN_FILE expands to ~/.skywork_token, which stores authentication material outside the skill directory. This is a real credential storage location in the user home directory.
High
Hidden file access
Token is stored in user home directory as ~/.skywork_token (JSON format).
The file described on this line is a hidden token cache, not ordinary application output. Access to this file can expose a Skywork bearer token.
High
Hidden file access
TOKEN_FILE = os.path.expanduser("~/.skywork_token")
The code defines a hidden home-directory token file used by later load and save operations. That creates persistent access to sensitive authentication data.
High
Python file write/append
with open(TOKEN_FILE, "w", encoding="utf-8") as f:
The code writes token data to the hidden token cache. I did not see permission hardening around this write, so the credential file may be created with default process permissions.
High
Python file write/append
with open(out_path, "w", encoding="utf-8") as f:
The file write uses a path derived from the raw query string on the previous line. Queries containing path separators or absolute paths can escape the intended temporary directory.
High
Hidden file in home directory
2. Cached token file `~/.skywork_token` → validate via API, if valid, use it
The skill documentation instructs use of a cached token file at ~/.skywork_token. That is a hidden credential file in the user home directory.
High
Hidden file access
2. Cached token file `~/.skywork_token` → validate via API, if valid, use it
The documented cached token is sensitive authentication material stored in a hidden file. Access to that file can expose the Skywork token.
Medium
Python environment access
API_BASE = os.environ.get("SKYWORK_API_BASE", "https://api.skywork.ai")
SKYWORK_API_BASE can change the API endpoint used for authentication checks. Because token validation sends credentials to that base URL, endpoint control is security-sensitive.
Medium
Python environment access
LOGIN_BASE_URL = os.environ.get("SKYWORK_WEB_BASE", "https://skywork.ai")
SKYWORK_WEB_BASE controls the login URL that may be opened or shown to the user. This is legitimate configuration, but it can steer users to a non-default login destination.
Medium
Python environment access
env_token = os.environ.get("SKYBOT_TOKEN", "")
The script reads SKYBOT_TOKEN directly from the environment. That is expected credential handling, but it is access to a sensitive secret.
Medium
Python environment access
env_token = os.environ.get("SKYBOT_TOKEN", "")
The command-line path reads SKYBOT_TOKEN and validates it with the API. This is legitimate authentication logic, but it handles a sensitive environment token.
Medium
Python HTTP libraries
with urllib.request.urlopen(req, timeout=30) as resp:
The script sends the search query and token-bearing request over the network. This is the core feature, but it exposes user queries and credentials to the configured API endpoint.
Medium
Python environment access
SKYWORK_GATEWAY_URL = os.environ.get("SKYWORK_GATEWAY_URL", "https://api-tools.skywork.ai/theme-gate
SKYWORK_GATEWAY_URL can replace the gateway that receives token-bearing search requests. This is useful configuration, but it is security-sensitive endpoint control.
Low
Hardcoded URL
API_BASE = os.environ.get("SKYWORK_API_BASE", "https://api.skywork.ai")
The script contacts the Skywork API for authentication and token validation. This is expected for the skill, but it is real outbound network behavior.
Low
Hardcoded URL
LOGIN_BASE_URL = os.environ.get("SKYWORK_WEB_BASE", "https://skywork.ai")
The script constructs login URLs against the Skywork web site. This is intended authentication behavior, but it sends the user to an external network service.
Low
Python HTTP libraries
req = urllib.request.Request(
The code builds an HTTP POST request to the Skywork gateway and includes the token header. This is intended behavior, but it is real outbound API access.
Low
Hardcoded URL
SKYWORK_GATEWAY_URL = os.environ.get("SKYWORK_GATEWAY_URL", "https://api-tools.skywork.ai/theme-gate
The default gateway points to api-tools.skywork.ai. A hardcoded external service is expected here, but it confirms the skill depends on outbound network access.
Risk Factors
⚙️ External commands (14)
🌐 Network access (6)
📁 Filesystem access (10)
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.
Copy report link
https://skillstore.io/skills/skyworkai-skywork-search/audits/5?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_reportMarkdown badge
[](https://skillstore.io/skills/skyworkai-skywork-search?utm_source=security_passport_badge)HTML badge
<a href="https://skillstore.io/skills/skyworkai-skywork-search?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/skyworkai-skywork-search/security.svg" alt="Skillstore security assessment" loading="lazy"></a>Embed card
<iframe src="https://skillstore.io/embed/skills/skyworkai-skywork-search.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
SkyworkAI. (2026). skywork-search security audit report (audit version 5) [Author version unspecified]. Skillstore. https://skillstore.io/skills/skyworkai-skywork-search/audits/5BibTeX citation
@techreport{skyworkai-skyworkai-skywork-search-2026,
author = {SkyworkAI},
title = {skywork-search security audit report (audit version 5)},
institution = {Skillstore},
year = {2026},
number = {5},
url = {https://skillstore.io/skills/skyworkai-skywork-search/audits/5},
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: "skywork-search security audit report (audit version 5)"
version: "unspecified"
type: report
authors:
- name: "SkyworkAI"
date-released: "2026-07-07"
url: "https://skillstore.io/skills/skyworkai-skywork-search/audits/5"
identifiers:
- type: other
value: "skillstore:skyworkai-skywork-search:audit:5"
description: "Skillstore immutable audit report identifier"
Skillstore Score
Why this score Evidence Confidence: Medium 45
Architecture
85
Maintainability
85
Content
69
Community
83
Spec Compliance
What You Can Build
Current Event Research
Check recent facts, source URLs, and snippets before writing a brief.
Document Source Gathering
Collect web references for reports, slides, or research summaries.
Technical Lookup
Find current release notes, documentation pages, or issue references.
Try These Prompts
Basic Search
Search for recent information about [topic] and summarize the most reliable sources.
Compare Sources
Search for [topic] from official sources, news coverage, and expert analysis.
Build A Brief
Research [question], cite source URLs, and separate confirmed facts from uncertain claims.
Multi-Step Research
Run focused searches for [topic], identify source conflicts, and produce a sourced decision brief.
Best Practices
- Use focused queries with time, location, product, or organization names.
- Open and read saved result files before answering the user.
- Cite source URLs when presenting factual claims.
Avoid
- Do not use broad single-word queries for complex research.
- Do not treat snippets as verified without checking source context.
- Do not include sensitive private data in search queries.
Frequently Asked Questions
What services does this skill use?
It uses the Skywork authentication and web search APIs.
Does it need credentials?
Yes. It needs SKYBOT_TOKEN, a cached token, or browser login.
Where are results saved?
The script saves result text files in a temporary directory and prints their paths.
How many queries can it run?
It accepts one to three queries per run.
Does it cite sources?
The result files include source URLs, and the assistant should cite them in answers.
Can it work without internet access?
No. It depends on Skywork API availability and network access.
Developer Details
Author
SkyworkAILicense
MIT
Skillstore revision
r1
Version notice
The author did not declare a version.
Ref
f93e9bb0daca99badb6a7e574b97737155d57cb3
Maintenance freshness
7/18/2026
Usage
12 downloads · 139 views
File structure