confluence-integration
Pass
Audited by Gen Agent Trust Hub on Jun 21, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: Read-only enforcement verified. The skill's scripts strictly use HTTP GET requests to the Confluence API and include explicit rules and instructions prohibiting write, update, or delete operations.
- [SAFE]: Credential management. The skill follows secure practices by instructing users to store Personal Access Tokens in a .env file (~/.env or project root). The load-env.sh script parses these local files safely using restrictive regex for environment variable keys.
- [SAFE]: Command execution safety. Subprocess calls to python3 for URL encoding are performed using static code templates. User-supplied parameters (CQL queries, expansion fields) are passed via standard input (bash herestrings), preventing command injection into the Python execution context.
- [SAFE]: Indirect Prompt Injection analysis. The skill ingests data from external Confluence pages, which represents a potential attack surface. However, this is the intended function of the tool and is performed without elevated permissions or unsafe prompt interpolation. * Ingestion points: confluence-get-page.sh, confluence-search.sh. * Boundary markers: Absent. * Capability inventory: Shell script execution, network requests (curl), and local file read access (.env). * Sanitization: None applied to API responses.
Audit Metadata