openai-security-threat-model
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes content from external repositories, which is an untrusted data source that could contain malicious instructions designed to influence the agent's behavior.
- Ingestion points: The
{repo_directory/path}variable inreferences/prompt-template.mdindicates the repository content to be ingested. - Boundary markers: Present; the System prompt in
references/prompt-template.mdenforces strict grounding and evidence-anchored reporting to prevent the agent from following instructions embedded in the analyzed code. - Capability inventory: The
Read,Grep,Glob,Write, andEdittools are used across the workflow defined inSKILL.mdto analyze files and generate the threat model report. - Sanitization: Present; the Security hygiene instructions in
references/prompt-template.mdmandate redacting secrets discovered during the analysis process.
Audit Metadata