ClawHub
Back to skill

Security audit

quantum-daily-tracker

Security checks for vulnerabilities and agentic risk

Overview

This skill is a coherent quantum-paper tracking workflow with disclosed RSS fetching and note/dashboard updates, though users should be aware of automatic scheduling and some plaintext RSS feeds.

Install this only if you want an automatic Chinese-language quantum literature tracker that fetches RSS feeds and updates files under quantum-tracker. Prefer HTTPS feed endpoints where available, review generated links before relying on them, and set clear limits for backfill runs so it does not create more notes than expected.

A.I.G

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 (1)

T09 · Insecure Skill Coding Practices

Warning
Location
SKILL.md:22
Finding

Unauthenticated RSS Transport Permits Content Tampering

Content
View full analysis

Vulnerability Details

File Location: SKILL.md:22-25, 51-56
Vulnerability Type: Plaintext HTTP transport for external RSS feeds
Risk Level: Medium

Vulnerable Code

markdown
| Quantum Physics               | http://export.arxiv.org/rss/quant-ph        |
| Optics                        | http://export.arxiv.org/rss/physics.optics  |
| Instrumentation and Detectors | http://export.arxiv.org/rss/physics.ins-det |
| Atomic Physics                | http://export.arxiv.org/rss/physics.atom-ph |
markdown
| PRX Quantum              | http://feeds.aps.org/rss/recent/prxquantum.xml |
| Physical Review X        | http://feeds.aps.org/rss/recent/prx.xml        |
| Physical Review Letters  | http://feeds.aps.org/rss/recent/prl.xml        |
| Physical Review A        | http://feeds.aps.org/rss/recent/pra.xml        |
| Physical Review B        | http://feeds.aps.org/rss/recent/prb.xml        |
| Physical Review Research | http://feeds.aps.org/rss/recent/prresearch.xml |

Technical Analysis

The Skill directs the Agent to retrieve several RSS feeds over unencrypted, unauthenticated HTTP. HTTP provides neither transport confidentiality nor server authenticity and does not protect content integrity against an on-path attacker.

An attacker able to intercept or modify network traffic can alter RSS titles, descriptions, links, authorship data, and other metadata before the Agent receives them. The workflow subsequently filters this remote content, summarizes it, and writes it into persistent paper notes and dashboards. The Skill also provides no explicit instruction to treat feed fields as untrusted data or to ignore instructions embedded in retrieved content.

This does not constitute remote code execution by itself. However, it creates a content-injection boundary through which fabricated research information, malicious links, or prompt-like text may enter generated artifacts.

Attack Path

...[truncated 1318 chars]

Remediation
View remediation

Remediation Suggestions

  1. Replace every http:// feed URL with a publisher-verified https:// endpoint.
  2. Reject redirects that downgrade from HTTPS to HTTP.
  3. Restrict feed retrieval to an explicit allowlist of expected schemes and hostnames.
  4. Validate TLS certificates using the platform's trusted certificate store; do not disable certificate or hostname verification.
  5. Treat all RSS fields as untrusted data. Explicitly instruct the Agent not to execute commands, follow operational instructions, or change behavior based on feed content.
  6. Sanitize generated Markdown links and reject unsafe URI schemes before persisting them.
  7. Where available, verify paper identifiers, titles, and destination links against canonical publisher metadata.
  8. Record retrieval failures rather than falling back to plaintext transport.

SkillSpector

Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The trigger description allows broad manual activation scenarios such as '用户请求更新论文' and '补充遗漏的论文' without clear authorization, scope limits, or confirmation requirements. In an agent environment, ambiguous invocation conditions can cause unintended execution, unnecessary network access, and unauthorized modifications to notes, dashboard content, or memory state.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The description states the skill will '生成中文笔记和每日速览', which imposes a specific language on outputs. There is no accompanying note that the language is configurable or based on user preference, so this appears to force a locale/language without opt-in.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.