Back to skill

Security audit

Daily Ai Brief Skill

Security checks for vulnerabilities and agentic risk

Overview

This AI news skill mostly does what it says, but its browser scraper disables Chromium sandboxing while visiting third-party pages, which creates a review-worthy safety risk.

Install only if you are comfortable with a network-heavy scraper that contacts many third-party sites and generates local reports from untrusted content. Run it as an unprivileged user or in a container, avoid opening generated Markdown in permissive renderers, and prefer a version that removes Chromium no-sandbox flags and pins dependencies.

Vulnerability Patterns
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
Findings (2)

T09 · Insecure Skill Coding Practices

Error
Location
src/modules/x_fetcher.py:124
Finding

Chromium Sandbox Disabled While Rendering Untrusted Third-Party Pages

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
src/modules/report_generator.py:265
Finding

Untrusted Feed Content Is Written to Markdown Without Context-Aware Escaping

Content
View full analysis
]*>.*?', '', content, flags=re.DOTALL | re.IGNORECASE ) content = re.sub(r'<[^>]+>', ' ', content) try: import html content = html.unescape(content) ``` URL validation only checks a string prefix: ```python if not (item.url.startswith("http://") or item.url.startswith("https://")): return False ``` ### Technical Analysis News titles, summaries, source data, and links originate from remote RSS feeds, APIs, scraped pages, Hacker News submissions, and third-party Nitter pages. These values cross an external trust boundary before being embedded in a generated Markdown document. The cleanup sequence is unsafe for output encoding because it removes literal tags first and then applies `html.unescape`. An encoded payload such as `<img src=x onerror=...>` contains no literal tag during the regular-expression removal stage. It is decoded into a literal HTML element afterward and is then written directly to Markdown. In addition, Markdown metacharacters are not escaped. A malicious title or summary can introduce headings, links, images, or raw HTML ...[truncated 2210 chars]
Remediation
View remediation
str: value = value.replace("\r", " ").replace("\n", " ") return re.sub(r"([\\`*_{}\[\]()#+\-.!|>])", r"\\\1", value) ``` 4. Parse URLs structurally and accept only `http` or `https` URLs with a non-empty hostname: ```python from urllib.parse import urlsplit, quote def safe_markdown_url(value: str) -> str: parsed = urlsplit(value.strip()) if parsed.scheme not in {"http", "https"} or not parsed.hostname: raise ValueError("Unsupported report URL") # Encode characters capable of terminating or restructuring Markdown links. return quote(value.strip(), safe=":/?#[]@!$&'*+,;=%") ``` 5. Reject control characters, embedded newlines, and malformed URLs. Consider restricting article links to the expected source domain when source provenance is important. 6. Configure the intended Markdown renderer to disable raw HTML. If reports are converted to HTML, apply a strict sanitizer and Content Security Policy to the resulting document. 7. Add regression tests using: - Entity-encoded HTML tags. - Raw HTML with event handlers. - Titles containing headings and image syntax. - URLs containing closing parentheses, newlines, and additional Markdown links. ]]>
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (32)

Tp4

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding

This finding indicates the skill has undeclared X/Twitter/Nitter scraping, browser automation via Playwright, and targeted social-media collection that are not transparently disclosed in the primary description. Undocumented browser automation and targeted scraping materially increase risk because they expand network reach, can interact with dynamic sites, and may collect or transmit more data than a user expects.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

This finding indicates the skill has undeclared X/Twitter/Nitter scraping, browser automation via Playwright, and targeted social-media collection that are not transparently disclosed in the primary description. Undocumented browser automation and targeted scraping materially increase risk because they expand network reach, can interact with dynamic sites, and may collect or transmit more data than a user expects.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The README content is presented in Chinese throughout, with no indication that the user can choose another language or that the locale restriction is intentional and documented as a regional requirement. Under the policy, forcing a specific language without user opt-in is a natural-language policy concern.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
89% confidence
Finding

The skill advertises capabilities that require file read/write and outbound network access, but it does not declare any tool scope or permissions boundaries. This can cause the runtime to grant broader access than users expect, reducing transparency and increasing the risk of unintended data access or exfiltration through third-party requests.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The skill describes automatic aggregation from many third-party sources but does not warn users that running it will generate outbound requests to numerous external domains. Lack of disclosure is dangerous because it can expose user IP/network metadata, trigger organization policy issues, and surprise users in restricted or privacy-sensitive environments.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The natural-language description 'AI Daily News Aggregator 主入口' mixes English with Chinese and indicates the skill is presented in Chinese by default. The file contains no natural-language indication that users can opt into another language or locale, which conflicts with the policy against forcing a specific language without user choice.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
87% confidence
Finding

The module docstring is written entirely in Chinese and provides no indication that users can choose another language or locale. The policy explicitly flags language or locale constraints when they are imposed without user opt-in or documented justification.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

This code uses Chinese-only natural-language strings in the module description and throughout generated report content, such as headings and labels, which effectively forces a specific language/locale. Under the policy, locale-specific behavior should either offer user opt-in or be clearly documented as intentionally region-specific.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
81% confidence
Finding

This file uses Chinese natural-language strings in the module description, docstrings, and log messages, which can impose a specific language without evidence of user opt-in. The policy explicitly flags language or locale constraints when the skill forces a specific language and does not offer a choice.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

This code explicitly sets locale='en-US', timezone_id='America/New_York', and Accept-Language: en-US,en;q=0.9 for all browsing sessions. Under the policy, forcing a specific language/locale without user opt-in is a natural-language policy violation unless the constraint is justified or configurable.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
85% confidence
Finding

This markdown file explains that the skill collects content from RSS, social media, APIs, and web scraping sources, and later states that reports are generated in the local reports/ directory. Because the README is the user-facing description for a skill that performs network access and writes output files, it should explicitly disclose these behaviors and any privacy or system-impact considerations.

Content

No source excerpt is available for this finding.

Unpinned Dependencies

Low
Category
Supply Chain
Confidence
97% confidence
Finding

The dependency is specified with a lower bound only, so builds may resolve to different versions over time and can unexpectedly pull in vulnerable or breaking releases. In a skill that aggregates external content and uses parsing/network libraries, this weakens supply-chain control and makes security posture hard to verify.

Content

Scanner excerpt · requirements.txt (reported line 1)May include surrounding context.

text
pyyaml>=6.0.3
feedparser>=6.0.12
requests>=2.33.1
beautifulsoup4>=4.12.0

Unverifiable Dependency: pyyaml has 8 known advisory(ies) (CVE-2019-20477 (Deserialization of Untrusted Data in PyYAML); CVE-2020-1747 (Improper Input Validation in PyYAML); CVE-2020-14343 (Improper Input Validation in PyYAML) +5 more), but the manifest does not pin a version, so it is unknown whether the installed release is affected

Low
Category
Supply Chain
Confidence
95% confidence
Finding

PyYAML has multiple historical advisories, and because the version is not pinned, there is no way to verify that installations avoid affected releases. If this skill parses YAML from untrusted or indirectly influenced sources, vulnerable versions could enable serious parser misuse such as unsafe deserialization or improper input handling.

Content

No source excerpt is available for this finding.

Unpinned Dependencies

Low
Category
Supply Chain
Confidence
97% confidence
Finding

The feedparser dependency is unpinned, which allows environment-dependent installs and makes it impossible to guarantee the same audited package version everywhere. Because this skill consumes external feeds, parser behavior and security fixes should be tightly controlled.

Content

Scanner excerpt · requirements.txt (reported line 2)May include surrounding context.

text
pyyaml>=6.0.3
feedparser>=6.0.12
requests>=2.33.1
beautifulsoup4>=4.12.0
lxml>=4.9.0

Unverifiable Dependency: feedparser has 10 known advisory(ies) (CVE-2011-1157 (feedparser Cross-site Scripting vulnerability); CVE-2009-5065 (feedparser Cross-site Scripting vulnerability); CVE-2011-1158 (feedparser Cross-site Scripting vulnerability) +7 more), but the manifest does not pin a version, so it is unknown whether the installed release is affected

Low
Category
Supply Chain
Confidence
93% confidence
Finding

feedparser has known advisories and the lack of a pinned version means the deployed package could still be an affected release. Since this skill ingests untrusted RSS/Atom content from external sources, parser vulnerabilities are more relevant in context.

Content

No source excerpt is available for this finding.

Unpinned Dependencies

Low
Category
Supply Chain
Confidence
98% confidence
Finding

An unpinned requests version means future installs may select different releases, including ones with unresolved security issues or changed TLS/redirect behavior. For a network-heavy news collection skill, that increases supply-chain and operational risk.

Content

Scanner excerpt · requirements.txt (reported line 3)May include surrounding context.

text
pyyaml>=6.0.3
feedparser>=6.0.12
requests>=2.33.1
beautifulsoup4>=4.12.0
lxml>=4.9.0
playwright>=1.40.0

Unverifiable Dependency: requests has 16 known advisory(ies) (CVE-2014-1830 (Exposure of Sensitive Information to an Unauthorized Actor in Requests); CVE-2024-47081 (Requests vulnerable to .netrc credentials leak via malicious URLs); CVE-2024-35195 (Requests `Session` object does not verify requests after making first request wi) +13 more), but the manifest does not pin a version, so it is unknown whether the installed release is affected

Low
Category
Supply Chain
Confidence
95% confidence
Finding

requests has several known advisories, and without exact version pinning it is impossible to confirm whether installs are patched. In a skill that performs outbound HTTP requests to many sources, issues involving redirects, credential handling, or TLS behavior can materially affect security.

Content

No source excerpt is available for this finding.

Unpinned Dependencies

Low
Category
Supply Chain
Confidence
95% confidence
Finding

beautifulsoup4 is unpinned, so deployments are not reproducible and may receive unreviewed upstream changes. While not directly proving exploitation, this creates avoidable supply-chain uncertainty in a scraper/parser workflow.

Content

Scanner excerpt · requirements.txt (reported line 4)May include surrounding context.

text
pyyaml>=6.0.3
feedparser>=6.0.12
requests>=2.33.1
beautifulsoup4>=4.12.0
lxml>=4.9.0
playwright>=1.40.0
aiohttp>=3.9.0

Unpinned Dependencies

Low
Category
Supply Chain
Confidence
98% confidence
Finding

lxml is a security-sensitive parser library and leaving it unpinned can expose deployments to parser-related vulnerabilities or inconsistent behavior across environments. In a skill that processes untrusted remote content, this is more concerning than a purely local utility dependency.

Content

Scanner excerpt · requirements.txt (reported line 5)May include surrounding context.

text
feedparser>=6.0.12
requests>=2.33.1
beautifulsoup4>=4.12.0
lxml>=4.9.0
playwright>=1.40.0
aiohttp>=3.9.0
fake-useragent>=1.4.0

Unverifiable Dependency: lxml has 14 known advisory(ies) (CVE-2021-43818 (lxml's HTML Cleaner allows crafted and SVG embedded scripts to pass through); CVE-2014-3146 (lxml Cross-site Scripting Via Control Characters); CVE-2021-28957 (lxml vulnerable to Cross-Site Scripting ) +11 more), but the manifest does not pin a version, so it is unknown whether the installed release is affected

Low
Category
Supply Chain
Confidence
97% confidence
Finding

lxml has a substantial advisory history, and leaving the version unpinned makes it impossible to verify whether the parser in use is safe. Because the skill parses untrusted web content, parser-related flaws can have a more meaningful impact here than in a closed-input application.

Content

No source excerpt is available for this finding.

Unpinned Dependencies

Low
Category
Supply Chain
Confidence
94% confidence
Finding

playwright is unpinned, which allows browser automation components to drift across installations and may introduce unexpected security or sandbox changes. Because this skill may render untrusted web pages, deterministic versioning is important.

Content

Scanner excerpt · requirements.txt (reported line 6)May include surrounding context.

text
requests>=2.33.1
beautifulsoup4>=4.12.0
lxml>=4.9.0
playwright>=1.40.0
aiohttp>=3.9.0
fake-useragent>=1.4.0

Unpinned Dependencies

Low
Category
Supply Chain
Confidence
97% confidence
Finding

aiohttp is unpinned, so network-facing behavior and security fixes cannot be verified for any given installation. For software that fetches remote data continuously, this undermines reproducibility and vulnerability management.

Content

Scanner excerpt · requirements.txt (reported line 7)May include surrounding context.

text
beautifulsoup4>=4.12.0
lxml>=4.9.0
playwright>=1.40.0
aiohttp>=3.9.0
fake-useragent>=1.4.0

Unverifiable Dependency: aiohttp has 16 known advisory(ies) (CVE-2024-52303 (aiohttp has a memory leak when middleware is enabled when requesting a resource ); CVE-2026-54279 (aiohttp: Host-Only Cookies Become Domain Cookies After CookieJar Persistence); CVE-2026-34514 (AIOHTTP has CRLF injection through multipart part content type header constructi) +13 more), but the manifest does not pin a version, so it is unknown whether the installed release is affected

Low
Category
Supply Chain
Confidence
94% confidence
Finding

aiohttp has known advisories, and the absence of an exact version prevents verification that the deployed release contains necessary fixes. While the actual exploitability depends on how the library is used, the unpinned network dependency creates avoidable uncertainty for an internet-facing skill.

Content

No source excerpt is available for this finding.

Unpinned Dependencies

Low
Category
Supply Chain
Confidence
93% confidence
Finding

fake-useragent is unpinned, allowing builds to consume varying upstream versions without review. Even if lower risk than core parser/network libraries, it still contributes to unnecessary supply-chain unpredictability.

Content

Scanner excerpt · requirements.txt (reported line 8)May include surrounding context.

text
lxml>=4.9.0
playwright>=1.40.0
aiohttp>=3.9.0
fake-useragent>=1.4.0

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
79% confidence
Finding

This manifest contains substantial natural-language content in Chinese alongside English identifiers and labels, but does not state whether the skill is intended specifically for Chinese-speaking users or whether other locales are supported. Under the language/locale policy rule, forcing or implicitly assuming a locale without user choice or justification can be a policy concern.

Content

No source excerpt is available for this finding.