atlassian-templates
Pass
Audited by Gen Agent Trust Hub on Nov 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted JSON specifications and incorporates their content into generated Confluence and Jira markup, creating a potential surface for indirect prompt injection.\n
- Ingestion points: Untrusted data enters the agent context through JSON files processed by
scripts/template_scaffolder.py,scripts/template_validator.py, andscripts/template_usage_analyzer.py(e.g.,assets/sample_template_spec.json).\n - Boundary markers: Absent. The scripts do not use delimiters or include instructions for the AI or downstream systems to ignore embedded commands within the template content.\n
- Capability inventory: The skill is designed to interact with corporate Atlassian instances via MCP tools such as
createConfluenceContentandcreateJiraIssue, which can be exploited if the generated markup contains malicious instructions.\n - Sanitization: Absent. The Python scripts perform direct string replacement and concatenation without escaping or validating the content of the input JSON fields.
Audit Metadata