T09 · Insecure Skill Coding Practices
Warning
- Location
- knowledge/domain.md:14
- Finding
- Plaintext HTTP Used for arXiv API Requests<![CDATA[ ## Vulnerability Details **File Location**: `knowledge/domain.md`, lines 14-17 **Vulnerability Type**: Plaintext transport for external API communication **Risk Level**: Medium ### Vulnerable Code ```md ### API Endpoint - **Base URL**: `http://export.arxiv.org/api/query` - **Method**: GET - **Rate Limit**: 1 request per 3 seconds (be respectful; no authentication required) ``` ### Technical Analysis The configured arXiv API endpoint uses unencrypted HTTP rather than HTTPS. Requests and responses therefore lack transport confidentiality, integrity, and authenticated server identity. An attacker with a privileged network position—such as a compromised gateway, malicious wireless access point, proxy operator, or upstream network adversary—could observe academic search queries or modify the returned Atom XML. Manipulated responses could contain fabricated bibliographic metadata, altered abstracts, or attacker-selected links that the Agent may treat as legitimate research results. ### Attack Path 1. The Agent follows the Skill instructions and sends a search request to `http://export.arxiv.org/api/query`. 2. A network-positioned attacker intercepts the plaintext HTTP connection. 3. The attacker modifies the API response, replacing paper metadata or PDF links with attacker-controlled content. 4. The Agent parses the manipulated response as trusted academic data. 5. Fabricated information or malicious links are incorporated into the Agent's recommendations and presented to the user. ### Impact Assessment This issue does not directly grant local code execution, filesystem access, or elevated system privileges. Its scope is limited to traffic sent through the insecure endpoint and content derived from its responses. Potential impact includes: - Disclosure of users' academic search queries. - Corruption of paper titles, authors, abstracts, publication status, and identifiers. - Injection of false research claims into literature summaries. - Redirecti ...[truncated 135 chars]
- Remediation
- <![CDATA[ ## Remediation Suggestions 1. Replace the plaintext endpoint with its HTTPS equivalent: ```md - **Base URL**: `https://export.arxiv.org/api/query` ``` 2. Require TLS certificate and hostname validation in the underlying HTTP client. Do not disable certificate verification. 3. Reject redirects that downgrade from HTTPS to HTTP. 4. Restrict redirects and response links to expected arXiv domains, such as `arxiv.org` and `export.arxiv.org`. 5. Validate returned records before presenting them: - Verify arXiv identifiers against the expected identifier format. - Require HTTPS for paper and PDF links. - Cross-check important metadata with Semantic Scholar or another independent source. 6. Treat titles, abstracts, and other externally retrieved text as untrusted data rather than Agent instructions. ]]>