T09 · Insecure Skill Coding Practices
- Location
scripts/nuclei_analyzer.py:139- Finding
Unsanitized Nuclei Fields Allow Markdown Report Content Injection
- Content
View full analysis
- Remediation
View remediation
Security audit
Security checks for vulnerabilities and agentic risk
This skill is a coherent Nuclei report generator with expected local file reading and report writing, but users should treat generated Markdown as untrusted scan-derived content.
Install only if you intend to process local Nuclei output files. Review the output path before running, avoid overwriting important files, and treat generated Markdown as untrusted if the scan data came from third parties or attacker-controlled targets.
scripts/nuclei_analyzer.py:139Unsanitized Nuclei Fields Allow Markdown Report Content Injection
The skill advertises behavior that reads input files, invokes local scripts, and writes reports to disk, but it does not declare any explicit tool scope or permission boundary. In an agent ecosystem, that ambiguity can allow broader-than-expected shell or file access during invocation, increasing the risk of unintended file reads/writes or command execution beyond the user’s intent.
The activation phrases like "analyze nuclei results" or "review scan" are broad enough to overlap with common security-analysis requests, which can cause the skill to trigger in contexts the user did not specifically intend. If auto-invoked, it could read local files or generate output as a side effect when the user was only asking for general advice or interpretation.
The workflow indicates that the skill reads a scan output file and generates a Markdown report, but the description does not clearly warn users about these filesystem side effects. Lack of disclosure reduces informed consent and can lead to unexpected handling of potentially sensitive scan data or unwanted report creation on disk.
Example prompts such as "Review scan findings and prioritize" are too generic to form a strong activation boundary, making accidental invocation more likely. While this is less severe than hidden code execution, it still increases the chance that the skill performs file-processing actions in response to ambiguous user requests.
This code performs a filesystem write by creating parent directories and writing a Markdown report, but there is no confirmation prompt or explicit user-facing warning about the write side effect beyond the generic CLI option description. For a code file, file writes are in scope for missing-warning review when there is no disclosure of the operation's effect on user data or filesystem state.