security-threat-model
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill's primary function is to ingest and analyze untrusted external data in the form of a source code repository. This architecture is susceptible to indirect prompt injection attacks, where an attacker could place malicious instructions within the project files (e.g., inside a README, code comments, or configuration documentation) to manipulate the agent's threat modeling logic or the final report output.
- Ingestion points: The agent reads arbitrary files from a user-provided repository path to extract system models and architecture details.
- Boundary markers: The skill relies on a system prompt template in
references/prompt-template.mdto define the agent's persona and objective, though it does not implement strict structural delimiters or escaping for the untrusted repository data it processes. - Capability inventory: The agent has the permission to read local files and write a Markdown-formatted report to the local file system.
- Sanitization: The skill includes explicit instructions to redact any secrets, tokens, or passwords found in the codebase to prevent accidental exposure in the generated threat model.
Audit Metadata